The influence of discrimination against migrants on the wage gap in Chinese urban labor market: A double selectivity approach

Abstract

This study explores the influence of discrimination against rural migrants on the wage gap in the Chinese urban labor market. This study empirically examines the extent to which discrimination against rural migrants influences the wage gap between the formal and informal sectors for the type of grouping between rural hukou holders and other hukou holders. For this purpose, we apply the Oaxaca-Blinder decomposition method, taking the formal/informal sector wage gap and the probability of entering each sector into the method (Brown model). In addition, we adopt a double-selectivity approach to examine the possibility of two selection biases. The results reveal that the unexplained components still significantly influence the wage gap, irrespective of whether double selectivity bias is considered. This indicates that our findings are inconsistent with those of existing studies, showing that the influence of discrimination against rural hukou holders on wage differentials has already vanished.

1.  Introduction

Poverty in China has declined dramatically since 1980; however, income inequality has expanded in the process of rapid economic development, although it has started to decline in recent times. A decomposition analysis of income inequality, wherein indicators of inequality are decomposed between and within rural and urban areas, reveals that intra-urban inequality continues to rise, while the inequality between rural and urban areas and intra-rural areas has declined (Xue, 2018; Yan, 2021).

One of the factors accelerating intra-urban inequality in China is the wage gap between rural migrants (“nouminko”) and urban residents. Discrimination against migrants from rural areas, based on the residence registration system (“hukou” system), has been considered to be a determinant of the wage gap.

However, Yan (2021), based on an empirical analysis, finds that the wage gap caused by discrimination against “nouminko” based on “hukou” system disappeared after 2010. Yan explains that this is because “hukou system” has lost its former role to segment the urban labor market, due to the institutional reform of this system by the Chinese government. It is possible that this reform reduced discrimination in small- and medium-sized cities (Zhang et al., 2019). However, Ma (2018), Siddique (2020), and Cheng et al. (2020) show evidence that discrimination against rural migrants still has a negative impact on rural migrants’ wages after 2010.

This study explores the influence of discrimination against rural hukou holders on the wage gap in the Chinese urban labor market based on Becker (1957)’s hypothesis. Becker (1957) posits that unexplained wage differentials can be attributed to wage discrimination. Following his argument, we consider the part of wage differentials unexplained by observable personal characteristics in the regression model as evidence of wage discrimination, while the remaining part, explained by observed characteristics, is attributed to productivity differences. Several studies have investigated this issue. However, controversy still exists regarding whether the wage gap caused by discrimination exists. While previous research has analyzed hukou-based wage discrimination in China from the viewpoint of occupational or industrial differences, wage disparities persist even within the same occupations or industries. This suggests that other factors, such as the nature of employment contracts, may significantly affect wage rates. Therefore, we distinguish sample between the formal and informal sectors based on the presence or absence of labor contracts. This method allows us to capture a crucial aspect of employment formality. By examining wage gaps between workers with and without formal labor contracts, we aim to shed new light on the mechanisms through which hukou status influences wage rates in both formal and informal labor markets. This study empirically examines the extent to which discrimination against rural hukou holders influences the wage gap between and within the formal and informal sectors for the type of grouping; between rural hukou holders vs. other hukou holders. For that purpose, we apply the Oaxaca-Blinder (hereafter O=B) decomposition method (Oaxaca, 1973; Blinder, 1973), taking the formal sector/informal sector wage gap and probability of entering each sector into the method (Brown model). In addition, we take into consideration the influences of double-selectivity bias both in sector allocation, and hukou status determination.

This study contributes to the literature in three ways. First, this study, to the best of our knowledge, is the first to apply O=B decomposition method with the Brown method to the wage gap between the group, considering the likelihood of working in the formal or informal sector, using data after 2010. Second, it adopts a double-selectivity approach based on the O=B decomposition method with Brown model to examine the influence of hukou status discrimination on the wage gap. This is the first study to apply the O=B decomposition method to a Brown model, taking account double-selectivity bias.

Third, this study is the first to investigate the wage discrimination against rural hukou holders, separating hukou holders into those who changed their hukou status from rural to urban and those whose original hukou status remains unchanged. In addition, this study is the first to examine the differences of wage discrimination against rural hukou holders between mega cities and the other cities.

According to the decomposition results for the case of rural versus other hukou holders, there still exists a significant influence of the unexplained components on the wage gap. These are inconsistent with the findings of existing studies, which show that the influence of discrimination against rural hukou holders on wage differentials does not exist anymore. The decomposition results also show that in the influences of the unexplained component of wage differentials, the influence within sectors is larger than that between sectors.

The remainder of this paper is organized as follows. Section 2 comprises a literature review. The third section explains the relevant methodology, data, and empirical strategies. The fourth section presents the estimation results. The final section concludes the paper.

2.  Literature review

Many studies have decomposition analysis to examine the determinants of the wage gap. These studies consider that the differentials of individual characteristics between rural migrants and urban residents and the discrimination against rural migrants are determinants of the wage gap.

Among those, most studies find that discrimination against migrants exerts significant influence on the wage gap (Chang and Zhao, 2016; Cheng et al., 2020; Deng, 2003; Deng, 2007; Fukui, 2022; Gravemeyer et al., 2011; Guo and Zhang, 2011; Ma, 2018; Meng, 2012; Meng and Zhang, 2001; Pakrashi and Frijiters, 2017; Siddique, 2020; Song, 2014; Song, 2016; Wang, 2003, 2005, 2007; Wang et al., 2015; Xue et al., 2014; Yao et al., 2008; Zhang and Guo, 2014; Zhu, 2016).

Conversely, some studies reveal that factors like differences on individual characteristics, type of industry, ownership type of enterprise, and presence or absence of a labor contract, are more important determinants of the wage gap as compared to the impact of discrimination, which has disappeared or diminished significantly (Alam and He, 2022; Demurger et al., 2009; Gagnon et al., 2014; Lee, 2012; Messinis, 2013; Yan, 2021; Zhang et al., 2016).

Among the studies finding significant impact of discrimination, Ma (2018), Meng and Zhang (2001), Yao et al. (2008), Wang (2007), and Zhang and Guo (2014) share the following common points: (i) They split households into two groups: urban households, whose head has an urban registered residence, and rural to urban migrant household, whose head has a rural registered residence which differs from the current urban residence; (ii) they apply the O=B decomposition method, using Brown model (Brown et al., 1980) which analyzes the influence of labor market segmentation by various sectors¹.

On the other hand, among studies that find negligibly small impacts of discrimination on the wage gap, Alam and He (2022), Demurger et al. (2009), Gagnon et al. (2014), Lee (2012), Messinis (2013), and Zhang et al. (2016) apply the O=B decomposition method, but not based on the Brown model. Of these studies, Gagnon et al. (2014) identify two groups with “hukou” status, rural hukou and other hukou, and investigate how the discrimination against rural hukou holders influences the probability of their entry into the formal sector².

However, most studies except Song (2016) do not consider the possibility of double selectivity bias caused by the correlation of the two selection processes.

The above literature review indicates that studies that apply the O=B decomposition method with Brown model draw a similar conclusion that the influence of discrimination against rural migrants on the wage gap are large without exception. However, they do not identify the groups with hukou status like as Gagnon et al. (2014) and Yan (2021) do, who find insignificant influences of discrimination against rural hukou holders. In addition, studies that analyze the effects of labor market segmentation, applying the O=B decomposition method and the Brown model, focus on occupational, employment-type and industry-sector segmentation. However, they do not analyze the influences of formal and informal sector segmentation, even though there is some empirical evidence that discrimination against migrants influences formal and informal sector segmentation and wage gap (Fukui, 2022; Gagnon et al., 2014). In addition, the studies do not take into consideration the influences of double selectivity bias. Finally, no study has investigated discrimination against rural hukou holders after separating cities into mega cities and other cities and urban hukou holders into those who changed their status from a rural hukou status and those whose status remained unchanged.

3.  Methodology, data, and empirical strategy

(1)  Methodology

This study empirically examines whether discrimination against rural migrants influences the wage gap between, and within the formal and informal sectors, for the type of groups: rural hukou holders versus other hukou holders. For this purpose, we apply the O=B decomposition method with Brown model and double selectivity approach, considering the formal sector/informal sector wage gap and probability of entry into each sector.

We first examine the discrimination against rural hukou holders by including all the samples in mega cities and other cities as well as separating the hukou holders into groups of those who had or had not changed their original status.

The wage function in O=B decomposition is formulated as follows;

  

$$\begin{array}{c}\hfill Ln{w}_i^g=X_i^g•{\beta }^g+u_i^g\end{array}$$(1)

Here, $i$ denotes the individual, $w$ is the wage rate, $X$ includes the individual characteristics, regional dummy variables affecting wage rate, and constant term. $\beta$ is the parameter, $u$ is the error term, and $g$ denotes group (rural hukou holders (r) or other hukou holders (o)). Other hukou holders include rural migrants who have transferred their hukou status from rural to urban areas.

The differences between the wage rates of the two groups is decomposed into composition effect (explained) and structure effect (unexplained);

  

$$\begin{array}{c}\overline{Ln{w}^o}-\overline{Ln{w}^r}\hfill \\ ={\beta }^o•(\overline{X^o}-\overline{X^r})+({\beta }^o-{\beta }^r)•\overline{X^r}\hfill \end{array}$$(2)

or

  

$$\begin{array}{c}\hfill ={\beta }^r•(\overline{X^o}-\overline{X^r})+({\beta }^o-{\beta }^r)•\overline{X^o}\end{array}$$(3)

where

  

$$\overline{X^o}=\frac{1}{n}{\sum }_i^n{X}_i^{o}i=1,2,\dots \dots ,n\text{,}$$

and

  

$$\overline{X^r}=\frac{1}{k}\sum _j^k{X}_j^{r}j=1,2,\dots \dots ,k\text{.}$$

The first and second terms on the right side denote the explained effect and the unexplained effect, respectively.

Comparing Eq. (2) with Eq. (3), we can see that the results of decomposition depend on the reference group, rural or other hukou holders, used to estimate counterfactuals. This is called the “index number problem”.

Various solutions have been proposed to cope with this problem (Neumark, 1988; Oaxaca and Ransom, 1994; Fortin, 2008). We employ Fortin’s methodology using parameters estimated by pooled data regression with a hukou dummy.

  

$$\begin{array}{ccc}\overline{Ln{w}^o}\hfill & -& \overline{Ln{w}^r}\hfill \\ & =\hfill & {\beta }^{o+r}(\overline{X^o}-\overline{X^r})+({\beta }^o-{\beta }^{o+r})•\overline{X^o}\hfill \\ \hfill & & +({\beta }^{o+r}-{\beta }^r)•\overline{X^r}\hfill \end{array}$$(4)

Here, ${\beta }^{o+r}$ is the parameters estimated by pooled data regression with hukou dummy.

Next, we use the Brown model to estimate how formal sector segmentation affects the wage gap between rural and other hukou holders.

Using the Brown model, we decompose the wage gap between rural and other hukou holders as follows:

  

$$\begin{array}{ccc}\overline{Ln{w}^o}\hfill & -\hfill & \overline{Ln{w}^r}\hfill \\ & =\hfill & \sum _h^{}{p}_h^r(\overline{X_h^o}-\overline{X_h^r}){\beta }_h^o\hfill \\ & & +\sum _h^{}{p}_h^r\overline{X_h^r}({\beta }_h^o-{\beta }_h^r)\hfill \\ & & +\sum _h^{}\overline{Ln{w}_h^o}(p_h^o-\widetilde{p_h^r})\hfill \\ & & +\sum _h^{}\overline{Ln{w}_h^o}(\widetilde{p_h^r}-p_h^r)\hfill \end{array}$$(5)

Here, $X_h^g$ represents individual characteristics and regional dummy variables of the laborer whose hukou type is g (r=rural hukou, o=other hukou) and entry sector is h (f=formal or i=informal), $p_h^g$ represents probability that the laborer who holds g type hukou enters h sector, $\widetilde{p_h^r}$ represents the counterfactual probability that rural hukou holders work in h sector when we assign the parameters estimated by using pooled sample.

If we consider Fortin’s methodology to cope with the “index number problem”, the wage gap equation can be reformulated as follows:

  

$$\begin{array}{ccc}\overline{Ln{w}^o}\hfill & -\hfill & \overline{Ln{w}^r}\hfill \\ & =\hfill & \sum _h^{}{p}_h^r(\overline{X_h^o}-\overline{X_h^r}){\beta }_h^{o+r}\hfill \\ & & +\sum _h^{}{p}_h^r\overline{X_h^o}({\beta }_h^o-{\beta }_h^{o+r})\hfill \\ & & +\sum _h^{}{p}_h^r\overline{X_h^r}({\beta }_h^{o+r}-{\beta }_h^r)\hfill \\ & & +\sum _h^{}\overline{Ln{w}_h^o}(p_h^o-\widetilde{p_h^r})\hfill \\ & & \begin{array}{c}\hfill +\sum _h^{}\overline{Ln{w}_h^o}(\widetilde{p_h^r}-p_h^r)\end{array}\hfill \end{array}$$(6)

Here, ${\beta }_h^{o+r}$ represents the parameters of individual characteristics of the laborer working in sector h. These were estimated based on pooled data.

The first term on the right hand side denotes the explained component in intra-sector wage differentials. The second and third terms on the right side denote the unexplained component in intra-sector wage differentials. The fourth term on the right side denotes the explained component in intra-sector wage differentials. The fifth term on the right side denotes the unexplained component in inter-sector wage differentials.

Tunali et al. (1980)’s model was applied (see also Song (2016)) to account for the influence of double selection bias. We estimate two selection equations using a bivariate probit model and estimate the following wage functions for laborers holding type g of hukou status and working in sector h: these include the selectivity correction terms.

  

$$\begin{array}{ccc}Ln{w}_{h,i}^g\hfill & =\hfill & X_{h,i}^g•{\beta }_h^g+{\lambda }_{h,i}^g•{\delta }_h^g\hfill \\ & & +{\lambda }_{g,i}^h•{\delta }_g^h+u_{h,i}^g\hfill \end{array}$$(7)

Here, ${\lambda }_{h,i}^h$ represent the selectivity correction terms generated from bivariate probit analysis. ${\delta }_h^g$ and ${\delta }_g^h$ represent the parameters.

Given the estimates from the regression of wage functions, we can reduce the estimated selectivity-corrected average wage differentials between rural and other hukou holders into the explained component in intra- and inter-sector differentials and the unexplained component in the intra- and intra-sector differentials in the same way as in equation (6).

  

$$\begin{array}{ccc}\overline{Ln{w}^o}\hfill & -\hfill & \overline{Ln{w}^r}\hfill \\ & -\hfill & \sum _h^{}(\overline{{\lambda }_h^o}•{\delta }_h^o+\overline{{\lambda }_o^h}•{\delta }_o^h-\overline{{\lambda }_h^r}•{\delta }_h^r+\overline{{\lambda }_r^h}•{\delta }_r^h)\hfill \\ & =\hfill & \sum _h^{}{p}_h^r(\overline{X_h^o}-\overline{X_h^r}){\beta }_h^{o+r}\hfill \\ & & +\sum _h^{}{p}_h^r\overline{X_h^o}({\beta }_h^o-{\beta }_h^{o+r})\hfill \\ & & +\sum _h^{}{p}_h^r\overline{X_h^r}({\beta }_h^{o+r}-{\beta }_h^r)\hfill \\ & & +\sum _h^{}\overline{Ln{w}_h^o}(p_h^o-\widetilde{p_h^r})\hfill \\ & & +\sum _h^{}\overline{Ln{w}_h^o}(\widetilde{p_h^r}-p_h^r)\hfill \end{array}$$(8)

In addition to the analysis based on entire dataset all over China, we compare the mega city where changing hukou status is more restricted to the other cities where hukou system institutional reforms have progressed relatively further for both those who changed hukou status and those who unchanged. This comparison allows us to examine the influence of hukou system’s institutional reform in China. We here consider the following four categories: megacity with changed hukou status, mega city with unchanged hukou status, other city with changed hukou status, and other city with unchanged hukou status

(2)  Data and empirical strategy

To examine the findings of Yan (2021) that the wage gap caused by discrimination against rural migrants to cities, based on “hukou” system has disappeared after 2010, we need to use the data surveyed after 2010. Therefore, we use the “urban residents personal data” and “migrants personal data” from the Chinese Household Income Project survey conducted in 2013 (CHIP 2013), incorporating the information about individual characteristics, formal-informal sectors, and wages of rural migrants and urban residents necessary for our analysis. We obtained these data from the CHIP conducted by the Economic Institute of Chinese Academy of Social Sciences and Beijing Normal University in 2014. These data have previously been used by Ma (2018) and Alam and He (2022).

The individuals with missing values used for empirical analysis and descriptive statistics are shown in Table 1.

Table 1. 

Descriptive statistics

	Rural hukou vs Other hukou
	Rural hukou		Other hukou
	Formal	Informal	Formal	Informal
Ln wage (log of wage rate)	2.72	2.32	2.85	2.34
Gender (Female=2, Male=1)	1.42	1.55	1.40	1.47
Educational years	11.07	9.52	13.01	10.95
Health condition (1. Excellent–5. Very poor)	1.58	1.86	1.82	1.94
Size of enterprize (1. Mimimum–7. Largest)	3.54	2.57	3.98	2.70
Number of other jobs	0.99	0.99	0.99	0.98
Experience (Working years)	14.33	5.66	22.09	7.85
Communist (Communist=1)	0.10	0.06	0.33	0.12
Type 2 (Foreign enterprises=1)	0.10	0.02	0.04	0.04
Type 3 (Domestic private enterprises=1)	0.51	0.57	0.17	0.51
Obs.	73	263	4,660	2,603

Source: Author's calculation based on CHIP 2013 data.

We explore the influence of discrimination against rural migrants on the wage gap in the Chinese urban labor market in the following steps.

First, we calculate the actual probabilities that the laborers of group g enter h ($p_h^g$), in which the dependent variable is a formal dummy (formal sector=1) and the independent variables are individual characteristics and regional dummy variables.

Second, we estimate the counterfactual probabilities, $\widetilde{p_h^r}$, applying bivariate probit analysis. For this, we assign the average values of the independent variables of rural hukou holders in the estimated parameters of bivariate probit functions using pooled sample.

Third, we estimate the parameters of the wage function for each group and sector (${\beta }_h^r$, ${\beta }_h^o$), and the parameters of the wage function for each sector including the selectivity correction terms are estimated based on pooled data (${\beta }_h^{o+r}$).

Finally, we assign all the probabilities and parameter values estimated as above in Eq. (8), and calculate the differentials explained and unexplained by individual characteristics in intra-sector wage differentials and inter-sector wage differentials.

4.  Results

Our analysis proceeded in two stages. First, we conducted a comprehensive analysis of the entire dataset. Second, we divided the dataset into subsets by city size and hukou changing status and conducted more focused analyses to investigate the influences of differences in the progress of hukou system’s institutional reform.

We start with the report using the entire data set. The actual entry probabilities ($p_h^g$) are shown in Table 2. From the results of bivariate probit analysis in the appendix A1 of online appendix, we calculate the counterfactual probabilities, $\widetilde{p_h^r}$, for Brown model (see also Table 2) and calculate selectivity correction terms and include them in the wage functions. The estimated parameters of the wage function for each group and sector (${\beta }_h^r$, ${\beta }_h^o$), and those of the wage function for each sector estimated based on pooled data (${\beta }_h^{o+r}$) are shown in the appendix A2 of online appendix.

Table 2. 

Probability of entry to formal or informal sector

Actual probabilities
$$p_f^r$$	0.2173
$$p_i^r$$	0.7827
$$p_f^o$$	0.6416
$$p_i^o$$	0.3584
Counterfactual probabilities
$$\widetilde{p_f^r}$$	0.4108
$$\widetilde{p_i^r}$$	0.5892

Source: Probabilities of counterfactual are calculated from the results of probit analysis shown in the appendix A1 of online appendix.

After we assign all the probabilities and parameter values estimated as above in Eq. (8), we calculate the differentials explained and unexplained by individual characteristics in intra-sector wage differentials and inter-sector wage differentials. The results are presented in Table 3. It shows decomposition results with double selectivity correction and without it. According to the table, if we do not consider double selectivity bias, 32% and 20% of the wage gap can be explained by intra-sector differentials of individual characteristics and inter-sector differentials, respectively.

Table 3. 

Decomposition results, using the Brown model

	Without selectivity correction		With double selectivity correction
	Actual value	Percentage (%)	Actual value	Percentage (%)
Total wage differentials	0.2546	100	0.2546	100
Selectivity corrected wage differentials			0.2220	87.22	100
Intra-sector differential
Explained	0.0819	32.16	0.0218	8.58	9.84
Unexplained	0.0927	36.43	0.1414	55.56	63.70
Inter-sector differential
Explained	0.0506	19.86	0.0320	12.55	14.39
Unexplained	0.0294	11.55	0.0268	10.53	12.07
Total explained differentials	0.1324	52.02	0.0538	21.13	24.23
Total unexplained differentials	0.1221	47.98	0.1682	66.09	75.77

Source: Calculated from the estimated probabilities and parameters.

Considering the double selectivity bias, 10% and 14% of the wage gap can be explained by the intra-sector differentials of individual characteristics and inter-sector differentials, respectively. However, 64% and 12% could not be explained by the intra-sector differentials of individual characteristics and inter-sector differentials, respectively. Compared to the results without double selectivity correction, wage discrimination within sectors is exacerbated by selectivity correction.

The results reveal that there still exists the significantly larger influence of the unexplained components on the wage gap, regardless of taking account of double selectivity bias. These are inconsistent with the findings of existing studies, which show that the influence of discrimination against rural hukou holders on the wage gap has already disappeared (Cai and Zhang, 2021; Gagnon et al., 2014; Yan, 2021), whereas these are consistent with existing studies, which show that the influences of discrimination on the wage gap between migrants and urban residents are too large to ignore (Ma, 2018; Meng and Zhang, 2001; Yao et al., 2008; Wang, 2007; Zhang and Guo, 2014).

Finally, regardless of the double selectivity correction, the influence of the unexplained component of the intra-sector differentials is greater than that of the explained component of the intra-sector differentials. In contrast, the influence of the explained component of the inter-sector component is greater than that of the unexplained component of the inter-sector differentials.

Next, we report the results based on the data from each subset. The results of considering double selectivity correction in the four separated categories are shown in Table 4. Here, the number of samples is 713 for hukou changed in mega city, 1297 for hukou unchanged in mega city, 2068 for hukou changed in other city, and 4257 for hukou unchanged in other city, respectively. According to this table, the influences of unexplained components of the intra-sector differentials on wage gap are significantly larger in all categories except for other city with changed hukou status. These findings are similar to those of the overall sample. In the category with other city and changed hukou, unexplained components in intra-sector differentials have a smaller influence on the wage gap than in the other categories. The reason that discrimination on rural hukou holders is smaller in this case could be interpreted as follows: In the cities other than the mega cities, the hukou system’s institutional reform has progressed relatively far. If discrimination is generated not only by emotional motivation but also via institutional regulations, the deregulation of the hukou system could reduce discrimination against rural hukou holders by urban hukou holders who have already changed their hukou status (from rural to urban) and experience a lower level of emotional discrimination against rural hukou holders.

Table 4. 

Decomposition results, using Brown method with double selectivity correction by city type and hukou change

	Mega City
	Hukou changed			Hukou unchanged
	Actual value	Percentage (%)		Actual value	Percentage (%)
Total wage differentials	0.1501	100.00		0.7009	100.00
Selectivity corrected wage differentials	0.0889	92.63	100.00	0.1344	80.82	100.00
Intra-sector differential
Explained	0.1545	12.79	13.81	0.1207	17.22	21.30
Unexplained	0.7814	64.71	69.85	0.2193	31.29	38.71
Inter-sector differential
Explained	0.1546	12.80	13.82	0.1078	15.37	19.02
Unexplained	0.1171	9.70	10.47	0.1188	16.95	20.97
Total explained differentials	0.3090	25.59	27.63	0.2284	32.59	40.32
Total unexplained differentials	0.8985	74.41	80.33	0.3381	48.23	59.68
	Other City
	Hukou changed			Hukou unchanged
	Actual value	Percentage (%)		Actual value	Percentage (%)
Total wage differentials	0.1619	100.00		0.1351	100.00
Selectivity corrected wage differentials	0.1284	79.29	100.00	0.0184	86.41	100.00
Intra-sector differential
Explained	0.0150	9.28	11.70	0.0200	14.81	17.14
Unexplained	0.0378	23.33	29.43	0.0554	41.00	47.44
Inter-sector differential
Explained	0.0412	25.45	32.10	0.0216	15.97	18.47
Unexplained	0.0345	21.34	26.91	0.0198	14.64	16.94
Total explained differentials	0.0562	34.72	43.80	0.0416	30.78	35.61
Total unexplained differentials	0.0723	44.67	56.34	0.0752	55.63	64.38

Source: Calculated from the estimated probabilities and parameters.

1) Mega city includes Beijing, and Guangzhou among five mega cities (Beijing, Shanghai, Guangzhou, Shenzhen, and Tianjin). The other city includes the cities other than five mega cities. In the case of mega cities, dummy variables were employed for Beijing. For other cities, dummy variables were used for regions 1, 2, 3, 5, and 6. Regional dummy variables are defined as follows: Region 1=1 if the region is Beijing, Tianjin, Hebei, Shanxi, or Inner Mongolia; Region 2=1 if the region is Liaoning, Jilin, or Heilongjiang; Region 3=1 if the region is Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, or Shandon; Region 4=1 if the region is Henan, Hubei, Hunan, Guangdong, Guangxi, or Hainan; Region 5=1 if the region is Chongqing, Sichuan, Guizhou, Yunnan, or Tibet; Region 6=1 if the region is Shanxi, Gansu, Qinghai, Ningxia, or Xinjiang=1. We used Region 4 for the base category.

2) As for the estimation results of wage functions, and bivariate probit analysis, we are prepared to provide them with the readers upon request.

5.  Conclusion

There is a controversy on whether the wage gap caused by discrimination against rural to urban migrants still exists.

Studies that apply the O=B decomposition method with the Brown model draw a similar conclusion that the influence of discrimination against rural migrants on the wage gap is large without exception. However, they do not identify the groups with hukou status as Gagnon et al. (2014) and Yan (2021) do, who find that the influence of discrimination against hukou holders is insignificant. In addition, the studies analyze the influence of labor market segmentation by applying the O=B decomposition method with the Brown model, which focuses on occupational, employment-type, and industry-sector segmentation. However, they do not analyze the influences of formal and informal sector segmentation, even though there is some empirical evidence that discrimination against migrants influences formal and informal sector segmentation and the wage gap.

This study empirically examines the extent to which discrimination against rural migrants influences the wage gap between and within the formal and informal sectors, for the case of grouping, “between rural hukou holders vs. other hukou holders”, applying the O=B decomposition method with Brown model. In addition, we consider the possibility of a double selectivity bias.

The key findings and implications derived from the analyses for both the entire data and the subset data can be summarized as follows.

First, the overall influence of discrimination against rural migrants on the wage gap is still large, not only in the case with double selectivity correction but also for the case without it. If we focus on the literature using data after 2010, it is consistent with those of Ma (2018) and Zhang and Guo (2014) for the case of rural migrants vs. urban residents. However, the case of rural hukou holders vs. other hukou holders is inconsistent with Yan (2021). This inconsistency suggests that the wage gap caused by discrimination against rural to urban migrants might not have disappeared even after 2010, although institutional reform of hukou system has been implemented by the Chinese government.

Second, the influence of the unexplained component of the wage gap within each sector is larger than the unexplained component of wage differentials between sectors regardless of double selectivity correction. This is mainly because of the large differentials between rural and other hukou holders in the influence of educational level on wages within the formal sector. Its effect accounts for 75% of the unexplained components of the intra-sector differentials. To show quantitative evidence, we give the breakdown of unexplained intra-formal sector differentials according to individual characteristics in eq. (8) (appendix A4 of online appendix). This finding is inconsistent with those of Gagnon et al. (2014) and Xue et al. (2014), which find that the influence of discrimination against rural hukou holders on intra-sector wage differentials is low.

The second finding implies that discrimination against rural migrants creates an unreasonable wage gap in the formal sector. This is explained by the fact that the differentials between rural and other hukou holders in the influence of individual characteristics on wages within the formal sector account for 86% of the unexplained intra-sector wage differentials (appendix A4 of online appendix). Ma (2018) recommends legislative action by the government, such as the “equal pay for equal work” regulation in Japan, to reduce such intra-sector wage gaps³. However, this might not be enough because this might contribute to a reduction in the wage gap between laborers with and without labor contracts, but not the intra-sector wage gap between rural migrants and others.

Third, we find that for the case of other city and changed hukou, the influences of unexplained components in intra-sector differentials on wage gap contribute to the wage gap less than they do in other cases. This could be interpreted as follows: In other cities, where the institutional reform of the hukou system has progressed relatively far, the deregulation of the hukou system could reduce some discrimination against rural hukou holders that is generated by institutional regulations. In addition, urban hukou holders who have changed their hukou status from rural to urban may hold less emotional discrimination against rural hukou holders.

These findings can contribute to the existing literature, although there are some limitations. In particular, we do not have any empirical evidence to support the interpretation of our third findings. We leave that for future research.

Acknowledgments

We thank Junichi Ito, Mitsuo Inada, and anonymous referees for their helpful comments on an earlier draft.

Notes

1  Meng and Zhang (2001) analyze occupational segmentation, and Wang (2007) analyzes employment type segmentation, and the others analyze industry sector segmentation. Cai and Zhang (2021), however, find the opposite results, although they also share the common points mentioned above.

2  Formal sector is defined as the sector in which the employee has permanent or long-term labor contracts with her employer. In CHIP 2013, type of labor contract was classified into ①permanent staff, ②long-term contract, ③short-term contract, ④without contract, and ⑤other. We define the enterprises using ① and ② type contracts with employees as the formal sector. Long-term contracts in the formal sector can be considered to be an internal labor market, which has gained wide recognition in labor economics. Based on the internal labor market theory, we assume that labor market segmentation by formal or informal sector influences the wage gap between rural hukou holders and other hukou holders because wages in the formal sector are higher than those of the informal sector and the hukou type is expected to affect the probability of entry into formal sector.

3  This regulation has been executed since April 2021, to remove unreasonable disparities in working conditions between non-regular employees and regular employees within the same working places.

References

•  Alam,  M. A. and  X.  He (2022) Is the Discrimination against Migrant Workers Tending toward Zero in Urban China?, International Studies of Economics 17: 65–88. http://doi.org/10.1002.ise3.4.
•  Becker,  G. (1957) The Economics of Discrimination, Chicago: University of Chicago Press.
•  Blinder,  A. S. (1973) Wage Discrimination: Reduced Form and Structural Estimation, Journal of Human Resources 8(4): 436–455. http://doi.org/10.2307/144855.
•  Brown,  R. S.,  M.  Moon, and  B. S.  Zoloth (1980) Incorporating occupational attainment in studies of male-female earnings differentials, Journal of Human Resources 15(1): 3–28. http://doi.org/10.2307/145344.
•  Cai,  S. and  X.  Zhang (2021) Anatomy of the Wage Gap between Local and Migrant Workers in Urban China: New Evidence from Matched Data. https://dx.doi.org/10.2139/ssrn.3933758.
•  Chang,  J. and  H.  Zhao (2016) The Impact of Work Unit’s Ownership on Wage Differentials between the Labor of Urban and Rural Household Registration, Chinese Economic Quarterly 15(2): 627–646 (in Chinese).
•  Cheng,  H.,  D.  Hu, and  H.  Li (2020) Wage Differential between Rural Migrant and Urban Workers in the People’s Republic of China, Asian Development Review 37: 43–60. https://doi.org/10.10.1162/adev_a_00140.
•  Demurger,  S.,  M.  Gurgand,  L.  Shi, and  Y.  Ximing (2009) Migrants as Second-class Workers in Urban China? A Decomposition Analysis, Journal of Comparative Economics 37: 610–628. https://doi.org/10.1016/j.jce.2009.04.008.
•  Deng,  Q. (2007) Earnings Differential between Urban Residents and Rural Migrants: Evidence from Oaxaca-Blinder and quantile Regression Decompositions, Chinese Journal of Population Science 2: 8–16 (in Chinese).
•  Fortin,  N. M. (2008) The Gender Wage Gap among Young Adults in the United States: The Importance of Money Versus People, Journal of Human Resources 43(4): 884–918. https://doi.org/10.3368/jhr.43.4.884.
•  Fukui,  S. (2022) The Urban Wage Gap and “hukou” System in China, The Economic Studies, The Journal of the Osaka University of Economics and Law 45(2): 37–44.
•  Gagnon,  J.,  T.  Xenogiani, and  C.  Xing (2014) Are Migrants Discriminated Against in Chinese Urban Labour Markets?, IZA Journal of Labor & Development 3: 17. http://doi.org/10.1186.2193-9020-3-17.
•  Gravemeyer,  S.,  G.  Gries, and  K. F.  Zimmerman (2011) Income Determination and Income Discrimination in Shenzhen, Uban Studies 48(7): 1457–1475. https://doi.org/10.1177/0042098010375317.
•  Guo,  F. and  S.  Zhang (2011) The Effect of Education Discrimination and Household Discrimination on the Wage Differentials between Urban Workers and Migrant Workers, Issues in Agricultural Economy 6: 35–42 (in Chinese).
•  Lee,  L. (2012) Decomposing Wage Differentials between Migrant Workers and Urban Workers, China Economic Review 23(2): 461–470. http://doi.org/10.1016/j.chieco.2012.03.004.
•  Ma,  X. (2018) Labor Market Segmentation by Industry Sectors and Wage Gaps between Migrants and Local Urban Residents in Urban China, China Economic Review 47: 96–115. https://doi.org/10.1016/j.chieco.2017.11.007.
•  Meng,  X. (2012) Labor Market Outcomes and Reforms in China, Journal of Economic Perspectives 26(4): 75–102. http://dx.doi.org/10.1257/jep.26.4.75.
•  Meng,  X. and  J.  Zhang (2001) The Two-Tier Labor Market in Urban China Occupational Segregation and Wage Differentials between Urban Residents and Rural Migrants in Shanghai, Journal of Comparative Economics 29: 485–504. https://doi.org/10.1006/jce.2001.1730.
•  Messinis,  G. (2013) Returns to Education and Urban-migrant Wage Differentials in China: IV Quantile Treatment Effects, China Economic Review 26: 39–55. https://dx.doi.org/10.1016/j.chieco.2013.006.
•  Neumark,  D. (1988) Employers’ Discriminatory Behavior and the Estimation of Wage Discrimination, Journal of Human Resources 23(3): 279–295. https://doi.org/10.2307/145830.
•  Oaxaca,  R. L. (1973) Male-female Wage Differentials in UrbanLabor Market, International Economic Review 14(3): 693–709. https://doi.org/10.2307/2525981.
•  Oaxaca,  R. L. and  M. R.  Ransom (1994) On Discrimination and the Decomposition of Wage Differentials, Journal of Econometrics 61(1): 5–21. https://doi.org/10.1016/0304-4076(94)90074-4.
•  Pakrashi,  D. and  P.  Frijters (2017) Migration and Discrimination in Urban China: A Decomposition Approach, Review of Income and Wealth 63(4): 821–840. https://doi.org/10.1111/roiw.12245?utm_source=ideas.
•  Siddique,  A. B. (2020) Identity-Based Earning Discrimination among Chinese People, IZA Journal of Development and Migration 11(1): 5. https://doi.org/10.2478/izajodm-2020-0005.
•  Song,  Y. (2014) What Should Economists know about the Current Chinese Hukou System?, China Economic Review 29: 200–212. https://doi.org/10.1016/j.chieco.2014.04.012.
•  Song,  Y. (2016) Hukou-based Labour Market Discrimination and Ownership Structure in Urban China, Urban Studies 53(8): 1657–1673. https://doi.org/10.1177/0042098015576861.
•  Tunali,  F.,  J. R.  Behrman, and  B. L.  Wolfe (1980) Identification, Estimation and Prediction Under Double Selection, Paper presented at the ASA meetings, Houston.
•  Wang,  M. (2003) Wage Differentials in the Transition Period: An Econometric Analysis on the Discrimination, The Journal of Quantitative & Technical Economics 5: 94–98.
•  Wang,  M. (2005) Urban Labor Market Discrimination against Migrants, Social Sciences in China 5: 36–46 (in Chinese).
•  Wang,  M. (2007) Changes of Discrimination Against Migrant Workers in China's Urban Labor Market, Chinese Labor Economics 1: 109–119 (in Chinese).
•  Wang,  H.,  F.  Guo, and  Z.  Cheng (2015) A Distributional Analysis of Wage Discrimination against Migrant Workers in China’s Urban Labour Market, Urban Studies 52: 2383–2403. https://doi.org/0042098014547367.
•  Xue,  J. (2018) Hinkon, Sitsugyo oyobi Syotoku Kakusa (Poverty, Unemployment, and Income Disparity), in Kajitani, K. and D. Fujii, eds., Gendai Chugoku Keizai Ron (Modern Chinese Economy) 2^nd ed., Kyoto: Minerva Publishing Co. Ltd., 84–93 (in Japanese).
•  Xue,  J.,  W.  Gao, and  L.  Guo (2014) Informal Employment and Its Effect on the Income Distribution in Urban China, China Economic Review 31: 84–93. http://dx.doi.org/10.1016/j.chieco.2014.07.012.
•  Yan,  S. (2021) Koseki Mibun no Tenkan to Syakai-keizai teki Kouka (Socio-economic Effects of Hukou Status and Change of Status from Rural Hukou), in Yan. S. Mikurodeita kara Miru Gendai Chugoku no Syakai to Keizai (Society and Economy of Current China: A Microdata Analysis), Tokyo: Keiso Publishing Co. Ltd., 91–122 (in Japanese).
•  Yao,  X.,  L.  Li, and  H.  Zhang (2008) Wage Discrimination and Occupational Segregation Against Rural Laborers‐Evidence of Zhejiang Province, China Management Magazine 227–239 (in Chinese).
•  Zhang,  D. and  Z.  Guo (2014) Industry Segmentation, Job Segregation and Wage Gaps by Registration System, Agriculture Technical Economics 5: 30–41 (in Chinese).
•  Zhang,  L.,  R. V.  Sharpe,  S.  Li, and  W. A.  Darity Jr. (2016) Wage Differentials between Urban and Rural-Urban Migrant Workers in China, China Economic Review 41; 222–233. http://dx.doi.org/10.1016/j.chieco.2016.10.004.
•  Zhang,  J.,  R.  Wang, and  C.  Lu (2019) A Quantitative Analysis of Hukou Reform in Chinese Cities: 2000–2016, Growth and Change 50: 201–221.
•  Zhu,  R. (2016) Wage Differentials between Urban Residents and Rural Migrants in Urban China During 2002–2007: A Distributional Analysis, China Economic Review 37: 2–14. https://doi.org/10.1016/j.chieco.2015.04.002.