Does Local Citizenship Still Matter? The Impact of Hukou Locality on the Employment of Relocated Households from the Perspective of Welfare Acquisition Cost

Abstract

Rural-to-urban resettlement is a widely used poverty alleviation strategy in China for fundamentally transforming poor farmers’ livelihoods, with roughly 263,000 rural poor relocated to urban communities in Liangshan Yi Autonomous Prefecture between 2016 and 2020. This dramatic development intervention in urbanization has unexpectedly resulted in the separation of hukou locality and residence. Considering that the government equally provides follow-up employment support policies to guarantee the citizenship rights for resettlers whether or not their hukou locality has transferred to urban communities, then, does the local citizenship still matter for employment? If so, how does local citizenship make a difference? What role does employment policy play in terms of the impact of local citizenship on employment? The answers to these questions are not yet clear. Based on a survey of 735 relocated households in the Liangshan Yi Ethnic Area, the Heckman sample selection model was used to empirically estimate the effect of local citizenship (hukou locality) on employment from the perspective of welfare acquisition cost, paying particular attention to the moderating effect of follow-up employment support policies. Our results show that (1) local citizenship can significantly increase the employed persons of a household by 0.279 units, prolong the working months by 2.297 units, and increase per capita wage by 0.885%. (2) Mechanism analysis shows that local citizenship affects relocated households’ employment by reducing welfare acquisition costs. (3) Moderating analysis shows that the follow-up employment support policies weakened the positive impact of local citizenship on employment, developing pro-poor jobs substitutes for the positive effect of local citizenship on the employed size; recommending job information substitutes for the positive effect of local citizenship on working months; and targeted labor exporting substitutes for the positive effect of local citizenship on per capita wage. This study provides new empirical evidence for understanding the relationship between hukou locality and employment consequences at a smaller scale and then provides theoretical reference and practical basis for the improvement of employment from the perspective of local citizenship.

1. Introduction

“Full and productive employment and decent work for all” is one of the Sustainable Development Goals (SDG8). Employment in cities is essential for the sustainable development of migrant families leaving their land. Compared to native urban residents, migrants from rural usually suffer lower employment rates, lower wages, and unstable occupation [1,2], especially for ethnic minority migrants [3,4]. According to a survey by the National Bureau of Statistics in 2023, the average monthly income of China’s rural migrant workers was CNY 4780 ¹, which was CNY 5278.167 and CNY 915 lower than the average monthly wages of CNY 10,058.167 and CNY 5695 for those employed in urban non-private and private sectors ², respectively. Improving migrants‘ employment contributes to the achievement of the Sustainable Development Goals.

During the 13th Five-year Plan period, China lifted 9.61 million rural registered poor people out of poverty by the Poverty Alleviation through Relocation (PAR) program, 78% of whom were relocated to centralized resettlements ³, and about 45% of the centralized population were relocated to urban communities in towns, cities or industrial parks. The Southwest regions have higher proportions of centralized resettlement. Liangshan Yi Autonomous Prefecture had resettled 353 thousand inhabitants living in deprived areas, 74.50% (263 thousand) of them moved to live in centralized resettlement communities in towns and cities ⁴. These communities are usually near county capital and prefecture-level cities, with a high concentration of employment demand, which is a high-risk area to prevent a large-scale return to poverty. Displacement and resettlement with transiting from a village- and farming-based livelihood to an urban one means a higher risk of unemployment and economic risk [5,6,7], which posed tremendous challenges to the integration of Yi minority resettlers into the urban labor market. The central and local governments aim to fundamentally transform the relocated households’ livelihoods by offering follow-up employment support policies in terms of guiding migrants to work out of the province, developing local pro-poor jobs, and improving employment services. Despite improvements in the employment rates of relocated migrants, employment stability remains poor and the quality of employment remains low. From the field survey in Liangshan Yi Autonomous Prefecture, the proportion that relocated households in urban communities are monitored for not returning to poverty because of unstable employment is 35.50%, which is 9.75 percentage points higher than other registered poor in rural.

The need to improve relocated households’ employment status has been the focus of academic attention, and the impact of citizenship on households’ employment decisions cannot be ignored [8,9]. In some developed countries, citizenship is often associated with the naturalization and integration of immigrants from economically less developed countries. The literature points out that granting citizenship could positively affect employment by permitting access to public sector jobs, as it provides a signal of a long-term commitment to remain, investing in human capital, and identifying with the host country. However, empirical findings reveal substantial ambiguity. Whereas some studies identify the faster access to citizenship, then higher employment rates, longer working hours, and more stable jobs for immigrants [10,11,12], others find that its impact is not universal and only benefits groups with higher potential capabilities [13,14], or even that there is no significant relationship [15,16].

In China, citizenship is usually associated with the world’s largest internal rural-to-urban mobility. Obtaining local hukou in urban communities, which is similar to a badge of citizenship in Western society, is the real standard by which to measure citizenship status, rights, and benefits [17]. Hukou status is classified into two parts: one by eligibility and one by registration place. The first classifies people into agricultural or non-agricultural hukou, which determines their entitlement to receive state subsidies and other privileges. The second classifies people into local or non-local hukou, which controls access to entitlements. In this case, hukou plays a critical role in the labor market. At the macro level, the reform of the hukou system relaxed the “entry barrier” of cities, promoted labor supply with higher skill [18], improved labor migration [19], reduced wage inequality [20], and promoted the economic growth of the non-farm sector [21], while some evidence showed that hukou reform showed no evidence in promoting short-term labor migration [22], and even strengthened the occupation segmentation between the local residents and newcomers [23]. At the micro level, most research focuses on the effect of urban local hukou conversion on employment, conflating the effect of hukou type and hukou locality. Compared to rural non-local hukou holders, urban local hukou holders are more likely to enter high-wage sectors [24,25], work in non-manual occupations [26], have an earnings advantage [27,28,29], and are less exposed to unemployment shocks [30]. A few studies identified the specific effect of hukou locality on residence-based welfare [31] and labor market vulnerability [32].

Up to now, we have provided a brief review of research on the relationship between citizenship and employment. The existing studies have not reached a consensus on the conclusion, and the existing understanding of the employment consequences after changing hukou locality across town units remains largely incomplete. In addition, considering that the government equally provides follow-up employment support policies to guarantee the citizenship rights for resettlers whether or not their hukou locality has transferred to urban communities, it is not clear that local citizenship still matters for employment among the group of ethnic minority relocated migrants. Then, our research questions are as follows: does local citizenship still matter for employment? If so, how does local citizenship make a difference? What roles does employment policy play in the impact of local citizenship on employment?

This paper aims to examine the effect of local citizenship on households’ employment based on survey data of 735 resettled households by applying the Heckman sample selection model, endogenous switching regression model, and instrumental variable methods to deal with endogeneity such as sample self-selection, reverse causation, etc. Compared to previous studies, the marginal contributions are as follows: First, we extend knowledge on the employment consequences of hukou locality at a smaller scale. Instead of taking hukou conversion as a mixed process of hukou type and locality, this study explores the employment consequences from the perspective of hukou locality. In addition, instead of focusing the hukou conversion scale on a prefecture or province, this study paid attention to the hukou locality transferring at a smaller scale such as towns. Second, the welfare acquisition cost was incorporated into the model, and the moderating effect of employment policies was explored. Lastly, we used a new dataset from medium, large, and mega-urban centralized resettlement communities ⁵. It collected a large-scale dataset from medium, large, and mega-urban resettlement communities in Liangshan Yi Autonomous Prefecture that covered 80% of the vulnerable Yi ethnic relocated population, addressing the shortcomings of the existing studies that view relocated migrants as a whole.

2. Background: Hukou System and Citizenship Under China’s Poverty Alleviation Through Relocation (PAR) Program

Hukou is a household registration system created by the Chinese central government in 1958 to manage the population and restrict labor mobility [33]. According to if a person is living in a rural or urban area, the hukou is divided into agricultural or non-agricultural types, and based on one’s birthplace or place of hukou registration, the hukou also was divided into local or non-local [34]. The hukou type (agricultural or non-agricultural) is closely linked to rural–urban divides in the type of entitlements, and the agricultural hukou holder access to less valuable benefits than urban citizens, while the hukou locality controls access to entitlements, and non-local hukou holders have no access to any benefit [32,35,36]. Citizenship entitlement to government services, such as public education, health insurance plans, and property rights, depends on the place where hukou is registered [33,37]. The central and local governments had committed to removing the correlation between hukou locality and public services, initiated China’s 2014 hukou reforms to abolish urban/rural distinction, and opened the hukou conversion in small and medium-sized cities. But rural migrants living and working in an urban city and away from their hukou registration place, a status of “hukou-living separation”, still face significant barriers to accessing welfare and public services in their place of residence [38].

The Poverty Alleviation through Relocation (PAR) program objectively results in a state of “hukou-living separation” in which the place of hukou registration and the place of residence of the relocated households do not coincide. The PAR program coincides with the local government’s plan to accelerate urbanization [39]. China had built 35 thousand resettlement communities and 2.66 million housing units between 2016 and 2020, with 9.61 million rural resettlers living in new homes and taken out of poverty [40]. About 78% of registered rural poor were resettled in centralized communities with superior supplementary facilities, and among those centralized resettlers, about 45% were relocated to county capitals, towns, cities, or industrial parks [41]. The rural poor who have relocated to urban resettlement communities are undergoing a dramatic transition of their identity from farmers to urban citizens in a relatively short time. Institutionally, full citizenship requires local hukou in the resettlement communities. The hukou institution for the relocated migrants is unclear around 2020, and some urban resettlement communities do not have a clear plan on whether the relocated households will be converted to urban household registration in the future. Few relocated households have transferred their hukou to the new community, considering that the rural hukou still binds the economic interests, such as land ownership, collective rents, and agricultural subsidies [39]. Since the local government has not improved the hukou management system and formulated social security programs for the relocated population, the original village cadres are unable to clearly explain the changes in land ownership, collective rents, agricultural subsidies, and social security. The relocated migrants anticipate damage to their original rights and loss of benefits caused by hukou conversion, so most of them chose to retain hukou in their original villages [42], which has resulted in them a “hukou-living separation” group whose hukou registration place and residence do not coincide. Since the medical care and pension insurance and agricultural subsidies of relocated families are tied to the place of registration of the hukou, relocated migrants need to go back to their original villages to complete the relevant procedures [43]. This hukou status has caused relocated households to float between the original village and the urban community, which has increased the cost of obtaining social welfare, and has also led to a reduced sense of belonging to the urban community, exacerbating the difficulties of identity transition for relocated households [41,42].

3. Theoretical Analysis and Research Hypotheses

3.1. The Effect of Local Citizenship on Employment

For the examination of the influence of local citizenship on households’ employment, the theories of naturalization and citizenship are the most influential ones [13,14,16]. Considering the transferring of hukou registration place among inner rural migrants in processes akin to international immigrant naturalization [44], we examined the employment effect of hukou locality based on the theory of naturalization and citizenship economics. China’s hukou institution categorizes residents into rural or urban, local or non-local of a particular location. The local urban residents in a municipality are favored in resource allocation compared to the non-local and rural migrants [45]. Since the relevant institutions governing local provision of public services, resource allocation, and a wide range of citizenship rights remain tied to the place of registration of the hukou [36,46], it is difficult for rural populations who migrate to cities other than their hukou-registered place to gain access to public services and rights equal to those of local urban residents [44]. Therefore, converting the hukou to a working city to acquire local citizenship is vitally important for the labor market outcomes of rural migrants when considering welfare acquisition.

The existing study has confirmed that improved social welfare such as public health insurance plans and pension plans is conducive to a better employment situation [47,48], while the hukou locality may have an impact on the cost of accessing social welfare, thus affecting relocated households’ employment. Although the hukou policies continue to be relaxed and access to public services has become more uniform and equal, especially in employment services, local schools, and housing security [39,49], distinctions remain in access to social security among migrants belonging to specific hukou locality. For example, Daochen County in Guizhou Province held a work meeting in 2019 to discuss the transfer of the Rural Minimum Living Standard Guarantee (Rural Dibao) program to the Urban Dibao program for poor relocated migrants, specifying the conditions that migrants must have, such as “holding an urban hukou, or holding a residence permit and having resided in a local urban district for more than one year”. The discussion of hukou locality and welfare acquisition costs among relocated migrants is limited around the year 2020. At that time, hukou conversion was directly related to the costs of welfare acquisition because the hukou management policy after relocation was not yet sound. Some research reported that relocated households who did not change their hukou to the resettlement communities still needed to return to their original villages to apply for social security such as the Dibao program, extreme poverty aid, and temporary assistance, as well as to enroll in healthcare insurance and pension insurance [42,43,49], which directly increased the cost of obtaining social welfare for the relocated households in their new communities.

Based on this, the following hypothesis was formulated.

Hypothesis 1 (H1). 

Hukou locality transferred to the resettlement community has a positive impact on employment.

Hypothesis 2 (H2). 

Hukou locality affects employment by reducing welfare acquisition costs.

3.2. Moderating Role of Follow-Up Employment Policies in Hukou Locality Affects Employment

The Chinese government implemented measures to support relocated migrants’ livelihoods, with stabilizing employment as the top priority. Given that relocated migrants have difficulties such as weak self-employment ability, few employment channels, and poor stability, the central and local governments offered follow-up employment support policies in terms of guiding migrants to work out of the province, developing local pro-poor jobs, and improving employment services [41]. The follow-up employment support policy reduces job search costs, increases the employment rate, and improves employment stability through a variety of means, such as skill training, developing pro-poor jobs, recommending job information, and providing cash subsidies for cross-provincial migration, which may weaken the positive effect of local citizenship on employment. Firstly, local governments launched skills training programs and certified skills qualifications, which may be alternatives to the effect of citizenship on employment by publicly investing in labor market-specific skills [10]. Secondly, developing pro-poor jobs by local governments could help to improve the employment rate, working hours, and annual earnings among marginalized migrants, thus weakening the positive effect of local citizenship on employment [10,11,12]. Third, the local government searches for job information on local enterprises to accurately match the relocated labor force with requirements and to reduce the job search costs of relocated families, which may weaken the positive effect of local citizenship on employment by developing local social networks [50,51,52]. Fourth, local governments organize the relocated labor force to targeted enterprises in developed cities to ensure that relocated migrants have access to stable, well-paid jobs, which may weaken the positive effect of local citizenship on standard employment and gaining access to the urban high-wage sector [24,53,54]. Fifth, local governments provide transportation subsidies and living allowances to relocated laborers who have worked steadily outside the province for three months or more, which may be an alternative to the positive effect of citizenship on working time [14]. These employment services cover all relocated households without setting threshold conditions based on hukou type or hukou locality, which is expected to weaken the positive impact of hukou on employment.

Accordingly, the following hypothesis was proposed.

Hypothesis 3 (H3). 

Follow-up employment support policy plays a moderating role in hukou locality affecting employment.

4. Materials and Methods

4.1. Data Source and Sampling

Liangshan Yi Autonomous Prefecture of Sichuan Province is the research area. Liangshan Yi Autonomous Prefecture was once a national poverty-stricken area and is regarded as a key area influencing Sichuan and even the whole country to win the battle against poverty alleviation. During the 13th Five-year Plan period, Sichuan Province completed the relocation of more than 379,000 households and 1.36 million rural poor out of poverty, ranking second in the country in terms of relocation scale. Liangshan Prefecture is not only the country’s largest Yi ethnic minority settlement area, but also was once the region with the largest number of rural poor people in Sichuan Province, and the population relocated out of poverty accounted for about 26% of the total size of the province. After relocation, Liangshan Prefecture became the region with the highest proportion of urban centralized resettlement communities in Sichuan province and the largest number of large-sized resettlement communities constructed. Thus, it is representative to choose Liangshan Yi Autonomous Prefecture as the sample area.

The data employed in this study are from the questionnaire survey carried out by the research group in Zhaojue County, Jinyang County, Butuo County, Yuexi County, Xide County, Ganluo County, Yanyuan County, and Leibo County, Liangshan Prefecture, Sichuan, China, in January 2022. The investigators are selected from postgraduates and doctoral students majoring in agriculture and forestry economic management and trained to ensure the quality of the survey data. The research combines stratified and random sampling methods to determine the survey samples. The specific steps are as follows: First, we stratified the different resettlement communities according to the size of the relocated population and location in each county; typical sampling was used to select 1~3 medium- and large-sized communities with more than 800 people, and 21 medium- and large-sized communities were selected. Second, we stratified the relocated households according to their annual income and randomly selected 40~50 relocated households for a questionnaire survey in each selected community. Thirdly, we collected data from the questionnaire using the Q&A method of questioning by the investigator—responses by the farmers—and data filling by the investigator. This survey mainly covers the basic characteristics of migrants, family size, employment information, and follow-up policy support. Finally, after the completion of data collection per day, a “triple-checking” procedure is required to ensure the quality of data, including “investigator self-checking”, “Team leader checking”, and then “Group teacher reviewing”. A total of 800 questionnaires were distributed, and after eliminating the samples with missing key information and containing extreme values, 735 valid samples were obtained, with a questionnaire validity rate of 91.87%.

4.2. Measures

4.2.1. Dependent Variable

The dependent variable is relocated households’ employment, which is analyzed in this study from the scale dimension, length dimension, and efficiency dimension, respectively. From the scale dimension, the employed size of households is mainly measured by the number of family members employed. Most households’ livelihoods after relocation have been transformed into non-farm livelihood activities, and paid jobs have become an important source of income for households. Referring to the definition of migrant labor ⁶ in the 2022 Migrant Worker Monitoring Survey Report, this study adopts the number of family members engaged in non-farming industries or out-of-hometown employment for more than six months in 2021 to measure employed size. From the length dimension, the labor supply of migrant households can be measured by working months. In this study, the cumulative working months that family members were engaged in non-agricultural industries or working outside the hometown in 2021 proxies the working months. From the efficiency dimension, the wages can measure the productivity of relocated migrants in the labor market, and this study proxies the labor productivity of migrants by the per capita wage of the family members employed in 2021, which is calculated by the ratio of the total wage income of household to the actual family population ⁷ in 2021.

4.2.2. Independent Variable

The independent variable is local citizenship, proxied by the hukou locality. Hukou locality is measured by “has your family transferred hukou to the resettlement community?”, where “yes” denotes the treated households assigned a value of 1, and “no” denotes the control households assigned a value of 0. The PAR program coincides with the local government’s plan to accelerate urbanization, as there is a clear preference for urban in the choice of resettlement location [39]. Most of the medium-, large-, and mega-sized resettlement communities are built in center towns, cities, or industrial parks [40,41], and are in different hukou administrative units from the original villages. Relocated households have the freedom to keep their hukou in their original villages or transfer it to the new resettlement communities. Therefore, this study argues that the hukou locality is a good proxy for migrants’ local citizenship in a new residence.

4.2.3. Control Variables

Referring to the studies of Zhang et al. [18] and Shi et al. [32], confounding variables in individual characteristics, household characteristics, economic conditions, and follow-up support policies were introduced as control variables. Firstly, individual characteristics such as age, age-squared, gender, education, and working experience reflect stronger adaptation capability and better human capital accumulation in urban cities, which is important for migrants’ economic integration and citizenship acquisition [55,56]. Secondly, household characteristics such as family size, labor force, and childcaring are related to higher livelihood capital, wider economic opportunities, better social support [57], and reallocation of family resources [58], and are thus directly associated with local citizenship and employment status. Thirdly, economic conditions such as household income, family risk, and Dibao (Minimum Living Standard Guarantee program) reflect the household economic foundation and risk, and poor economic condition indicates more difficulty in accessing local citizenship and better occupation [59]. Fourthly, the follow-up support policies such as microcredit for industrial development and employment policies directly affect the permanent residence in urban communities and lead to improved occupational well-being [6,60].

Table 1. Variable definition and descriptive statistics.

Variables	Definition	Mean	Mean by Group
			Treated Group	Control Group
Dependent variables
Employed size	Number of household members being employed by non-farm or leaving hometown for work 6 months or more in 2021.	1.034	1.220	0.986
Working months	Number of months that family members employed by non-farm or out for work in 2021.	10.653	12.027	10.301
Per capita wage	Gross household wage income in 2021 divided by actual population, in logarithms.	7.723	7.815	7.699
Independent variable
Citizenship	1 if the hukou registration of the respondent is at the resettlement communities, and 0 otherwise.	0.204	/	/
Mechanism variable
Welfare acquisition cost	1 if the household has to return to the original village for social security procedure, and 0 otherwise.	0.098	0.020	0.118
Moderating variables
Skill training	Number of family members participating in the skill training program in 2021.	1.088	0.840	1.152
Pro-poor jobs	Number of family members accessing pro-poor jobs in 2021.	0.297	0.300	0.296
Job information	1 if the household accesses job information through local government, and 0 otherwise.	0.275	0.293	0.270
Labor export	1 if the household labor exported by a government-organized program, and 0 otherwise.	0.029	0.047	0.024
Transport subsidy	1 if the household access transport subsidy for cross-province migration, and 0 otherwise.	0.035	0.073	0.026
Control variables
Age	Age of the head of household (in years).	47.906	48.953	47.638
Age-squared	Age square of the head of household divided by 100.	25.085	26.306	24.772
Gender	1 if the household head is male, 0 otherwise.	0.834	0.773	0.850
Education	1 if the head of household is illiteracy, 2 if primary school education, 3 if middle school education, 4 if high school or vocational school education, 5 if college degree and above.	1.737	1.673	1.754
Working experience	1 if the household has a family member employed out of county for 6 months or more in 2020, and 0 otherwise.	0.665	0.680	0.662
Family size	Number of people residing in the household.	5.005	4.920	5.027
Labor force	Number of people aged 16-59 years who can participate in the labor market.	1.910	1.827	1.932
Childcaring	The number of preschool-aged children and school-aged children in the household.	1.952	1.987	1.944
Household income	Total household income in 2020 divided by the actual population, in logarithms.	9.168	9.207	9.158
Family risk	Assigned 1 if the household is monitored for preventing a return to poverty because of illness, disability, accidents, unstable or sudden drop in income, unstable employment, lack of labor force, and weighted average with equal weights (1/6).	0.148	0.186	0.138
Dibao	Number of household members accessed the Minimum Living Standard Guarantee program in 2020.	1.721	1.387	1.807
Microcredit	1 if the household accesses microcredit for the targeted poor, and 0 otherwise.	0.374	0.373	0.374
Employment policy	1 if the household has a family member out for work through government-led employment projects, access local job information, transport subsidies, and living allowances, respectively, and the cumulative score is calculated.	0.352	0.433	0.332

shows the results of the descriptive analysis of the variables. Around 20.41% of the relocated households moved their hukou to the resettlement community. There is a significant difference in the employment situation between the treated group and the control group. The employed size, working months, the per capita wage, and the per capita income of the treated group are higher than that of the control group, which reflected the systematic differences in development opportunities between urban communities and rural villages. In addition, the treated households have lower costs in welfare acquisition. As a whole, 9.8% of relocated households need to return to their original villages in 2020 when they enroll in health insurance and pension insurance, as well as social security procedures such as the Minimum Living Standard Guarantee program, extreme poverty aid, and temporary assistance. Among relocated households whose hukou has been relocated, the proportion is 2%, while among relocated households whose hukou has not been relocated, the proportion is 11.8%. This phenomenon reflected differences in the cost of accessing social welfare in different hukou localities.

4.3. Methods

4.3.1. Heckman Sample Selection Model

Relocated households’ employment outcomes after changing hukou locality can be divided into two stages. The first stage is the households’ choice on whether to transfer hukou locality from rural to urban communities, which is the result of a cost–benefit analysis of their integration into the city; the second stage is, if they choose to change hukou locality, how their employment is affected. In this case, we can only observe households’ employment outcomes when they choose to change hukou locality in the first stage. There would be a potential bias if we ignored such a sample selection bias by estimating households’ employment directly. Therefore, this paper employs the Heckman sample selection model to estimate the effect of local citizenship on employment. The basic model is as follows:

$$y_i={\beta }_0+{\beta }_1{CIZ}_i+\gamma X_i+{\epsilon }_i$$

(1)

In Equation (1), $y_i$ denotes the employment of the relocated household $i$, including the employed size working months, and per capita wage. $X_i$ denotes a series of confounding variables, ${CIZ}_i$ denotes the local citizenship, and assigns a value of 1 if the households have moved their hukou locality from the original village to the resettlement community, and 0 otherwise. $CIZ$ is not exogenous, whether to change hukou locality is a self-selected behavior of migrants, and there are several unobservable factors simultaneously affect the decision to change hukou locality and its employment outcomes. Then, the decision equation is as follows:

$${CIZ}_i^{*}=\gamma Z_i+{\mu }_i, {CIZ}_i=1 ({CIZ}_i^{*}>0)$$

(2)

In Equation (2), ${CIZ}_i^{*}$ denotes the net benefit of changing hukou locality, which is the basis for hukou migration decisions, ${\mu }_i$ is a random perturbation term, and $Z$ denotes an exogenous variable that has an impact on relocated households’ hukou decisions. Based on the exclusivity constraint requirement, $Z$ contains one or several variables other than $X$, which has strong explanatory power for $CIZ$, but has no direct effect on $y$.

Different hukou locality decisions correspond to different employment outcomes. The potential employment outcomes of relocated households under the two decisions are shown below as follows:

$$y_i=\left\{\begin{array}{c}c if {CIZ}_i=1\\ 0 if {CIZ}_i=0\end{array}\right.$$

(3)

Since the perturbation terms $\epsilon$ and $\mu$ are correlated, then $E\left(\epsilon |CIZ\right)\ne 0, \mathrm{i}.\mathrm{e}.,\mathrm{t}\mathrm{h}\mathrm{e}$ OLS estimate ${\beta }_1$ of Equation (1) is biased. The idea behind the estimation of the Heckman sample selection model is to control for this bias by performing a probit regression of Equation (2) using the MLE method and constructing the inverse of Mill’s ratio (IMR, λ); then, with the selection bias corrected, a consistent estimate of ${\beta }_1$ in Equation (4) is obtained.

$$y_i={\beta }_0+{\beta }_1{CIZ}_i+\gamma X_i+{\rho \sigma \lambda }_i{+\omega }_i$$

(4)

On this basis, by constructing endogenous switching regression to conduct counterfactual estimation through Equations (2) and (3), the average treatment effect (ATT) of employment outcomes for households who changed hukou locality can be obtained through Equation (5), as well as the average treatment effect (ATU) of employment outcomes for households who did not change hukou locality in Equation (6).

$$ATT=E\left(y_{1i}|{CIZ}_i=1\right)-E\left(y_{0i}|{CIZ}_i=1\right)$$

(5)

$$ATU=E\left(y_{1i}|{CIZ}_i=0\right)-E\left(y_{0i}|{CIZ}_i=0\right)$$

(6)

4.3.2. Moderating Effect Model

Based on theoretical analysis, a moderating effect model is designed to test the moderating role of follow-up support policy on the impact of local citizenship on employment. The specific model is as follows:

$$y_i={\beta }_0+{\beta }_1{CIZ}_i+{\beta }_2{M}_i+{\beta }_3{CIZ}_i\times M_i+\gamma X_i+{\epsilon }_i$$

(7)

where $y_i$, ${CIZ}_i$, $X_i$, and ${\epsilon }_i$ in Equation (7) is the same as Equation (1). $M_i$ stands for the moderate variables, namely follow-up support policy. ${\beta }_3$ is the interactive coefficient of the hukou locality and follow-up support policy. The estimation of the entire model was implemented through Stata 18.

5. Results

5.1. Heckman Sample Selection Model Results

To overcome the sample self-selection problem, the Heckman sample selection model was used to analyze the determinants of the conversion of hukou locality, and, subsequently, factors affecting employment outcomes. The probit model is used in the first stage, where the dependent variable is the hukou locality. This variable is in binary form and assigns the value 1 if the household had a registered hukou at the resettlement communities and 0 otherwise. The OLS model was applied in the second stage, where the employed size, working months, and per capita wage are specified as the dependent variables.

Column (1) of

Table 2. The impact of citizenship on employment of relocated households.

	(1)	(2)	(3)	(4)
Variables	Selection Equation	Outcome Equation
		Employed Size	Working Months	Per Capita Wage
Hukou locality		0.279 ***	2.297 *	0.885 **
		(0.093)	(1.282)	(0.347)
Age	0.001	0.019 *	0.217 ***	0.116 **
	(0.001)	(0.010)	(0.072)	(0.043)
Age-squared		−0.016	−0.196 ***	−0.136 ***
		(0.010)	(0.068)	(0.038)
Gender	−0.046	0.050	−0.012	0.104
	(0.033)	(0.081)	(0.511)	(0.240)
Education	−0.008	0.059	0.677 **	0.164
	(0.022)	(0.045)	(0.301)	(0.127)
Working experience	0.006	0.725 ***	5.071 ***	1.603 ***
	(0.030)	(0.075)	(0.612)	(0.278)
Family size	−0.005	0.110 ***	0.987 ***	0.315 ***
	(0.015)	(0.033)	(0.284)	(0.084)
Labor force	−0.012	0.105 **	0.920 **	0.393 ***
	(0.019)	(0.038)	(0.332)	(0.115)
Childcaring	0.014	−0.055	−0.512	−0.140 *
	(0.015)	(0.037)	(0.322)	(0.078)
Household income	0.014	0.265 ***	2.288 ***	0.552 *
	(0.033)	(0.062)	(0.607)	(0.286)
Family risk	0.054	−0.299 **	−3.572 **	−1.419 *
	(0.077)	(0.136)	(1.313)	(0.715)
Dibao	−0.013 **	−0.033 ***	−0.166	−0.061
	(0.006)	(0.010)	(0.115)	(0.046)
Microcredit		−0.025	−0.004	−0.146
		(0.061)	(0.581)	(0.217)
Employment policy		0.245 ***	4.078 ***	0.420 ***
		(0.057)	(0.724)	(0.142)
Hukou conversion in the community	0.649 ***
	(0.057)
Residence length	0.011
	(0.013)
County GDP	−0.000 **
	(0.000)
IMR		−0.047	−0.597	−0.500 **
		(0.068)	(0.844)	(0.231)
Constant		−3.192 ***	−27.116 ***	−2.934
		(0.652)	(6.237)	(2.723)
N	735	735	735	735
R2	0.228	0.476	0.467	0.383

Note: (1) Robust standard errors in parentheses. (2) The hukou conversion in the community (was calculated by the average rate of hukou transferring among other households in the same community), residence length (was calculated by the years resident in the resettlement community), and county GDP are exclusion constraint variables and are only included in the selection Equation. (3) *** p < 0.01, ** p < 0.05, * p < 0.1. The same as below.

presents the probit regression results. The Dibao program is significant and negative (p < 0.05), which indicates that migrant families with poor economic conditions are reluctant to give up their rural hukou for reasons of survival rationality and to cope with future uncertainty. In addition, the average proportion of hukou conversion in the community was positively and significantly (p < 0.01) related to the decision to change the hukou locality of certain households. This implies that hukou conversion may be a collective action, and the decision to change hukou is affected by the demonstration effect among peers.

Columns (2)~(4) in Table 2 present the OLS results. The hukou locality coefficients have positive and significant effects on the employed size (p < 0.01), working months (p < 0.1), and per capita wage (p < 0.05), respectively, which indicates that households changing their hukou registration place were more likely gaining employment advantages and premiums. This is consistent with the conclusions of Zhang [31] and Shi et al. [32]. The possible reason is as follows: the hukou locality is still associated with access to entitlements, since the relevant institutions governing local provision of public services, resource allocation, and a wide range of citizenship rights remain tied to the place of registration of the hukou [36,46], procedures and processes for rural populations who migrate to cities other than their hukou-registered place to access public services and rights may differ from those of local urban residents [44].

From the results of control variables, the age of the household head shows an inverted U-shaped relationship with employment outcomes, with household employment size, working months, and per capita wage showing a tendency to first increase and then decrease as the age increases. Education (p < 0.05), working experience (p < 0.01), family size (p < 0.01), labor force (p < 0.05 at least), and household income in the previous year (p < 0.01) all have significant positive effects on employment outcomes; these results are consistent with several previous studies by Peters et al. [56], Wang and Fan [55], and Shi et al. [32]. The possible reasons are as follows: the higher the level of education, working experience, and more of a labor force, the higher the level of household livelihood capital, and the higher the competitiveness in the labor market. And the higher the family size and household income, the better social support and economic conditions for searching for better occupations. Childcaring (p < 0.1), family risk (p < 0.1), and Dibao (p < 0.01) both have significant negative effects on employment outcomes, which are in line with the conclusions of Yang and Tang [61], Aleksandrova et al. [62], and Xu and Carraro [59], respectively. The possible reasons are as follows: First, more preschool-aged and school-aged children mean heavier caregiving tasks for family labor in a relocated household, which has a “locking in” effect on the free flow of family labor, thus lowering employment rate, working time, and even income. Second, the higher probability the household is monitored to prevent a return to poverty and more members access to the Dibao program, reflecting the worse personal and household circumstances, which determine long-term unemployment. Employment policy (p < 0.01) has a significant positive effect on household employment outcomes; this is in line with the conclusions in the impact of employment projects on economic integration of Tang et al. [6].

5.2. Robust Check

5.2.1. Endogenous Switching Regression Model

Social capital is vitally important in the employment outcomes of relocated households in the city. We use an endogenous switching regression model to address the endogeneity problem caused by omitting important variables such as social capital.

Table 3. Endogenous switching regression treatment effects.

		Hukou Locality Decision		ATT	ATU
Variables		Change Hukou	No Change in Hukou
Employed size	Treated group	1.220	0.980	0.240 ***
	Control group	1.211	0.986		0.225 ***
Working months	Treated group	12.027	10.167	1.860 ***
	Control group	12.652	10.301		2.351 ***
Per capita wage	Treated group	7.816	7.201	0.615 ***
	Control group	8.930	7.699		1.231 ***

Note: The treated group is households that have changed their hukou locality from village to resettlement communities. The control group is households that did not change their hukou locality. *** p < 0.01.

presents the expected value of employment outcomes under observed and unobserved scenarios. First, the employed size dropped from 1.220 to 0.980 if the treated group household did not change their hukou locality, and the employed size increased from 0.986 to 1.211 if the control group household changed their hukou locality. The average treatment effect in the treated (ATT) of hukou locality on employed size is 0.240 (p < 0.01), and the average treatment effect in the untreated (ATU) is 0.225 (p < 0.01). Second, the working months decreased from 12.027 to 10.167 if the treated group household did not change their hukou locality, and the working months increased from 10.301 to 12.652 if the control group household changed their hukou locality. The average treatment effect in the treated (ATT) of the hukou locality on working months is 1.860 (p < 0.01), and the average treatment effect in the untreated (ATU) is 2.351 (p < 0.01). Third, the per capita wage decreased from 7.816 to 7.201 if the household had not changed the hukou locality, and the per capita wage increased from 7.699 to 8.930 if the control group household had changed their hukou locality. The average treatment effect in the treated (ATT) of hukou locality on per capita wage is 0.615 (p < 0.01), and the average treatment effect in the untreated (ATU) is 1.231 (p < 0.01).

5.2.2. Instrumental Variable Approach

To overcome endogeneity from reverse causality between local citizenship and employment, we exploited an instrumental variable (IV) based on the average rate of hukou transferring among other households in the same community. We argue that hukou registration policies are often announced at the county level and implemented by the community. Relocated households living in the same resettlement community enjoy the same information on hukou policies and are likely to obtain social evaluations on hukou transferring from their neighboring peers, which can influence their hukou transferring decisions. The employment of relocated households is the result of intra-household division of labor and joint decision-making and is not affected by the hukou status of external households.

We proceed with the two-stage least squares (2SLS) estimation of the impact of local citizenship on employment and report the results in

Table 4. IV estimation on the impact of local citizenship on employment.

	(1) Employed Size		(2) Working Months		(3) Per Capita Wage
Variables	First Stage	Second Stage	First Stage	Second Stage	First Stage	Second Stage
Hukou locality		0.305 **		1.965		0.816 *
		(0.145)		(1.290)		(0.481)
IV	0.887 ***		0.887 ***		0.887 ***
	(0.074)		(0.074)		(0.074)
Weak identification	182.40		182.40		182.40
N	735

Note: robust standard errors in parentheses, Cragg–Donald F-stat was reported in weak identification, *** p < 0.01, ** p < 0.05, * p < 0.1.

. Column (1)~(3) displays a strong first-stage relationship between hukou locality and the average rate of hukou transferring among other households in the same community (IV) in our sample. The greater the average rate of hukou transferring among other households, the greater the likelihood of changing the hukou locality to the new community. The first stage F-statistic of the instrumental variable is 182.40, which is by far greater than the conventional critical value (i.e., 10), suggesting no weak instrumental variable problem. The second stage in Columns (1) and (3) shows that hukou locality has a positive and significant effect on households’ employed size (b = 0.305, p < 0.05) and per capita wage (b = 0.816, p < 0.1). Column (2) shows that hukou locality has a positive effect on working months, but the coefficient is statistically insignificant. This partly verifies research hypothesis H1.

5.3. Mechanism Analysis: Based on Welfare Acquisition Cost

Following the theoretical analysis, the interaction term of hukou locality and welfare acquisition cost was used to further explore the mechanism of local citizenship affecting employment. We assessed the cost of welfare access to households based on two questions. The first question is “Does the household need to back to the original village to complete procedures related to social security such as the Dibao program, extreme poverty aid, temporary relief, etc.?”. The second question is “Does the household need to go back to the original village to complete procedures regarding health insurance and pension insurance?”. If the answer to any question is YES, it is assigned a value of 1, which means that the cost of access welfare is high; otherwise, it is 0, which means that the cost is low. The results are shown in

Table 5. Mechanism of the welfare acquisition cost.

	(1)	(2)	(3)
Variables	Employed Size	Working Months	Per Capita Wage
Hukou locality	0.279 ***	2.296 *	0.886 **
	(0.082)	(1.250)	(0.337)
Welfare acquisition cost	−0.306 **	−2.161 *	−0.529 *
	(0.127)	(1.072)	(0.280)
Hukou locality × welfare acquisition cost	0.886 ***	8.019 ***	0.415
	(0.221)	(1.901)	(0.406)
Control variables	YES	YES	YES
imr	−0.079	−0.853	−0.542 **
	(0.061)	(0.825)	(0.241)
Constant	−3.184 ***	−26.979 ***	−2.967
	(0.660)	(6.453)	(2.648)
N	735	735	735
R2	0.487	0.475	0.385

Note: robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

. The coefficients on welfare acquisition cost on employed size (b = −0.306, p < 0.05), working months (b = −2.161, p < 0.1), and per capita wage (b = −0.529, p < 0.1) are all significant and negative. The possible reason is that the higher the welfare acquisition cost, the worse it is for unhealthy workers to accumulate health capital [47], blocking the increased free flow of young migrant labor force [48], and thus, the worse it is in improving working conditions for the relocated households. The coefficients of the interaction terms of hukou locality and welfare acquisition cost on employed size and working months are all significantly positive at the 1% level, and the coefficients on per capita wage are positive but not statistically significant. This suggests that hukou locality has a positive impact on employment outcomes by reducing welfare acquisition costs. This verifies research hypothesis H2.

5.4. Moderating Role of Follow-Up Employment Support Policies

In this section, skill training, pro-poor jobs, job information, labor export, and transport subsidy are selected as proxy variables for employment support policies. Skill training is measured by the “number of family members participating in the skill training program in 2021”. Pro-poor jobs are measured by the “number of family members accessing pro-poor jobs in 2021”. Job information is assigned a 1 if the household accesses job information through local government, and 0 otherwise. Labor export is assigned a 1 if the household labor is exported by a government-organized program, and 0 otherwise. Transport subsidy is assigned a 1 if the household access transport subsidy for cross-provincial migration, and 0 otherwise.

Table 6. Moderating effect analysis of follow-up employment policies.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	Employed Size	Working Months	Per Capita Wage	Employed Size	Working Months	Per Capita Wage
Hukou locality	0.223	2.153	1.400 *	0.422 ***	3.287 **	0.896
	(0.191)	(1.664)	(0.721)	(0.158)	(1.295)	(0.569)
Skill training	0.012	0.350	0.222 **
	(0.037)	(0.355)	(0.109)
Hukou locality × skill training	0.185	0.925	−0.434
	(0.247)	(1.546)	(0.491)
Pro-poor jobs				0.343 ***	6.013 ***	0.239
				(0.107)	(1.043)	(0.203)
Hukou locality × pro-poor jobs				−0.541 *	−5.289	−0.369
				(0.323)	(3.478)	(0.752)
	(7)	(8)	(9)	(10)	(11)	(12)
	Employed Size	Working Months	Per Capita Wage	Employed Size	Working Months	Per Capita Wage
Hukou locality	0.410 **	3.350 **	0.947*	0.316 **	2.390 *	0.936 *
	(0.159)	(1.303)	(0.575)	(0.148)	(1.361)	(0.494)
Job information	0.465 ***	8.027 ***	0.606 ***
	(0.096)	(0.951)	(0.216)
Hukou locality × job information	−0.468	−5.851 *	−0.552
	(0.320)	(3.245)	(0.808)
Labor export				0.246	3.332	1.641 **
				(0.422)	(4.011)	(0.784)
Hukou locality × labor export				0.099	−5.042	−4.074 *
				(0.832)	(10.193)	(2.184)
	(13)	(14)	(15)
	Employed Size	Working Months	Per Capita Wage
Hukou locality	0.321 **	2.210	0.844 *
	(0.149)	(1.372)	(0.493)
Transport subsidy	0.108	−3.632	0.733
	(0.412)	(3.852)	(0.852)
Hukou locality × transport subsidy	−0.006	7.476	−0.817
	(0.741)	(6.806)	(1.645)

Note: robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the regression results. First, Columns (1) to (3) show that the interaction terms of hukou locality and skill training have insignificant effects on employed size, working months, and per capita wage. Second, Columns (4) to (6) show that the coefficients of the interaction term between hukou locality and pro-poor jobs on employed size are negative and significant at the 10% level, and the coefficients of the effects on working months and wage are both negative but not statistically significant. Considering that the coefficients of pro-poor jobs on employment outcomes are all significantly positive, we argue that developing pro-poor jobs weakened the positive effect of the hukou locality on employed size. Third, Columns (7) to (9) show that the coefficients on the interaction terms of hukou locality and job information are both negative but not statistically significant for employed size and wage, and the coefficients on working months are negative and significant at a 10% level. Considering the effect of job information on employment outcomes are all significantly positive, we argue that job information weakened the positive effect of hukou locality on length of employment. Fourth, Columns (10) to (12) show that the interaction term between hukou locality and labor export has insignificant coefficients on both employed size and working months and a significantly negative coefficient on per capita wage. Combined with the significantly positive coefficient of the effect of labor export on per capita wage, we argue that government-organized labor export weakened the positive effect of hukou locality on wage. Fifth, Columns (13) to (15) show that the interaction terms of hukou locality and transport subsidy have negative coefficients on employed size and per capita wage, and positive coefficients on working months, but none of them is statistically significant. The results verify research hypothesis H3.

In sum, we argue that follow-up employment support policies can replace the employment promotion role of hukou locality by developing local pro-poor jobs, precisely matching employment supply and demand information, and organizing targeted labor export to developed cities, thus weakening the positive impact of hukou locality on the employment outcomes of relocated households. Specifically, developing pro-poor jobs can provide employment opportunities for relocated migrants having difficulty finding jobs through market channels, thus weakening the positive impact of local citizenship on households’ employment size. The provision of job information by the government can compensate for the social network disadvantages of relocated migrants, thus weakening the positive effect of local citizenship on working months. Targeted labor exporting helps relocated migrants to enter high-paying, high-security workplaces, compensating for the inherent disadvantage of migrants’ access to labor welfare and thus weakening the positive impact of local citizenship on per capita wage.

6. Conclusions and Policy Implications

6.1. Conclusions

In China, the hukou locality was once closely associated with the basic rights and public services of local citizenship. The Chinese government has long been committed to eliminating the urban–rural divide in the hukou system and has now fully liberalized hukou registration restrictions in small- and medium-sized cities. The rural poor who have been resettled in urban communities have full discretion in changing their hukou locality, and the government has provided follow-up employment support policies to fundamentally transform their livelihoods whether or not their hukou locality has changed. Then, does changing the hukou locality to access local citizenship still matter for relocated households’ employment? If so, how does local citizenship make a difference? What role does employment policy play in the impact of local citizenship on employment?

To answer the questions mentioned, based on the survey data of 735 relocated households in 18 urban centralized resettlement communities across 8 counties in Liangshan Yi Autonomous Prefecture, we use the Heckman sample selection model to test the effect of local citizenship (hukou locality) on employment outcomes and its mechanism, and the moderating effects model to test the role of employment policies in the impact of local citizenship on employment.

We find that local citizenship still matters for households’ employment, with hukou locality transferred to urban communities, households’ employed size increased by 0.279 units, working months increased by 2.297 months, and per capita wage increased by 0.885%. Mechanism analysis shows that local citizenship significantly affects employment by reducing welfare acquisition costs in social security. Further analysis shows that the impact of local citizenship on employment is somewhat weakened in an intensive employment support policy environment, developing pro-poor jobs, matching job information and targeted labor exporting weakened the positive impact of local citizenship on households’ employment size, working months, and per capita wage, respectively.

Our study also has limitations, and future research could be expanded in at least two ways. First, future research could expand the sample to other provinces such as Guizhou, Yunnan, and Guangxi, as these are also areas where the size of the relocated poor population is larger and where there is a higher concentration of ethnic minorities. Second, considering that each region has different hukou management programs for relocated migrants—for example, Guizhou tends to implement unified hukou conversion at the community level, while Sichuan tends to support voluntary decentralized hukou conversion at the household level—future research could deepen the impacts of the different hukou management on the employment of relocated migrants. Thirdly, further research can delve into the long-term effect of local citizenship on the economic integration and social status of relocated families, to deepen the understanding of the long-term effects of citizenship policies.

6.2. Policy Implications

The above findings suggest that follow-up support measures should focus on reducing the costs of welfare acquisition and strengthening employment services.

First, optimizing community “one-stop” public services, opening digital social security processing channels, and reducing the welfare acquisition cost in social security for relocated migrants. Second, local governments should take into account the need for urbanization, developing and expanding pro-poor jobs in public service, and priority should be given to poverty-stricken laborers who are “unable to leave home, have no jobs, and unable to lift themselves out of poverty” and who can perform the work. Third, local governments should build public information networks by constructing county platforms, and community employment service stations to integrate relocated labor forces and accelerate the flow of employment information. Last, local governments should take advantage of the East–West Labor Collaboration Strategy to communicate closely with developed cities and enterprises in the eastern region on employment demand information and expand the scale of targeted labor exporting for relocated laborers.