Impact of Social Capital on Chinese Migrant Workers' Poverty during COVID-19: The Mediation of Social Protection

Abstract

Attention to and investigation of migrant workers’ poverty level in the COVID-19 environment are vital for understanding their living conditions. It is also critical to explore the effect of social capital and protection on migrant workers’ poverty alleviation in the post-COVID era. Using an online survey, this study examines the relationship between migrant workers’ social capital and poverty alleviation at the aggregate and dimensional levels from the multidimensional poverty perspective in the COVID-19 environment. We find that, in the COVID-19 environment, migrant workers’ social capital in the cities is notable for mitigating their poverty, and accessible social protection can weaken this relationship at the aggregate level. At the dimensional level, the result shows that migrant workers’ poverty alleviation depends on their social network and reciprocal connections in the cities, not social trust. However, migrant workers’ accessible social protections can mediate the relationship between the three dimensions of social capital and poverty. Our findings provide new evidence for the good and dark sides of the social capital and poverty alleviation relationship. Our result also shows that social protection can mitigate the influence of unequal social capital on poverty to achieve a more balanced result. The findings suggest that governments should design more inclusive but targeted social protection policies for migrant workers to decrease the effect of unequal social capital and so increase the impact on poverty alleviation.
JEL classification: D10; I32; J61

Full text

Page 1/27
Impact of Social Capital on Chinese Migrant Workers' Poverty
during COVID-19: The Mediation of Social Protection
Chen Chen (  chenchen_zp1226@163.com )
Lincoln University
Christopher Gan
Lincoln University
Research Article
Keywords: social capital, poverty alleviation, multidimensional poverty, social protection, migrant workers
Posted Date: December 5th, 2022
DOI: https://doi.org/10.21203/rs.3.rs-2327254/v1
License:   This work is licensed under a Creative Commons Attribution 4.0 International License.  Read Full License

Page 2/27
Abstract
Attention to and investigation of migrant workers’ poverty level in the COVID-19 environment are vital for understanding their
living conditions. It is also critical to explore the effect of social capital and protection on migrant workers’ poverty alleviation in
the post-COVID era. Using an online survey, this study examines the relationship between migrant workers’ social capital and
poverty alleviation at the aggregate and dimensional levels from the multidimensional poverty perspective in the COVID-19
environment. We nd that, in the COVID-19 environment, migrant workers’ social capital in the cities is notable for mitigating their
poverty, and accessible social protection can weaken this relationship at the aggregate level. At the dimensional level, the result
shows that migrant workers’ poverty alleviation depends on their social network and reciprocal connections in the cities, not
social trust. However, migrant workers’ accessible social protections can mediate the relationship between the three dimensions
of social capital and poverty. Our ndings provide new evidence for the good and dark sides of the social capital and poverty
alleviation relationship. Our result also shows that social protection can mitigate the inuence of unequal social capital on
poverty to achieve a more balanced result. The ndings suggest that governments should design more inclusive but targeted
social protection policies for migrant workers to decrease the effect of unequal social capital and so increase the impact on
poverty alleviation.
JEL classication: D10; I32; J61
1. Introduction
Households were severely affected by the COVID-19 pandemic, but not all households can shift their income-generating activities
to those not impacted by the pandemic (Mahmud and Riley, 2022), especially migrant workers with constrained working skills.
There are an estimated 292million domestic migrant workers in China, comprising over one-third of the working population
(China Labour Bulletin, 2021). Because of institutional restrictions on the household registration and health system, migrant
workers are not entitled to the same protections as local workers, especially migrant workers who are uncontracted and have low
skills (Wang, 2022). A considerable number of migrant workers cannot access social services, such as the health system,
education system and unemployment security (Li et al., 2022). Those institutional constraints weaken their ability to cope with
the economic shocks from COVID-19, exposing them to the edge of poverty.
Numerous studies have shown that social capital can relieve poverty by improving welfare and providing greater access to
resources for the poor (Kamarni et al., 2019; Pham and Mukhopadhaya, 2021; Tahmasebi and Askaribezayeh, 2021). However,
social capital is not naturally given but is established and rearmed through symbolic and material practices and transactions
(Das, 2004). Additionally, social capital has exclusive attributes that could oust someone who violates the norms or beliefs in
networks (Seferiadis et al., 2015). Migrant workers, especially rst-generation migrant workers who do not have migrant networks
in cities, need to build social capital through institutionalizing relationships (Portes, 2009), forming bounded solidarity and
creating value introjection, reciprocity exchanges, or enforceable trust (McGrath, 2010). However, the investment cost of social
capital could worsen migrant workers’ poverty rather than relieve it. It is still unclear whether social capital can help mitigate
migrant workers' poverty. In addition, poverty is multidimensional, though most studies focus on the relationship between social
capital and unidimensional poverty, such as using income (Heizmann and Böhnke, 2016; Lukasiewicz et al., 2019) or expenditure
(Islam & Alam, 2018; Ma et al., 2020) to measure poverty. Some scholars recently explored the relationship between social capital
and poverty from the multidimensional poverty perspective (Cheng et al., 2022; Osei and Zhuang, 2020; Pham and
Mukhopadhaya, 2021), and the results are inclusive. For example, Osei and Zhuang (2020) found that social capital can alleviate
multidimensional poverty from the perspective of women's entrepreneurship and social innovation. However, Cheng et al. (2022)
show that high social capital brings about elite capture, impeding the poor from obtaining beneciary quotas in poverty targeting.
Furthermore, social protection provides a series of formal and informal measures for people’s well-being (Boccagni, 2011) that are
critical in reducing poverty and inequality (Heizmann and Böhnke, 2016). However, migrant workers lack the necessary social
protection in cities (Li et al., 2022; Morales, 2016), limiting their access to resources that can improve their living state. In the
COVID-19 environment, it is unclear how much social protection migrant workers can access and how the social protection they
gain helps them reduce their poverty.

Page 3/27
This study examines how social capital alleviates migrant workers’ poverty from the multidimensional poverty perspective during
the COVID-19 crisis and the role of social protection. Using an online survey questionnaire, we collect 610 valid responses.
Specically, we explore the relationship between migrant workers’ social capital and poverty and the mediation effect of social
protection at the aggregate level. We split social capital into three dimensions: social networks, reciprocity, and social trust, to
explore how these dimensions inuence migrant workers’ poverty at the dimensional level, and how social protection mediates
their effects. The Two-Stage Residual Inclusion (2SRI) approach and Two-Stage Least Squares (2SLS) are used to address the
endogeneity problem. The results show that migrant workers’ poverty can be alleviated by their social capital in the cities.
Dimensionally, migrant workers’ social network and reciprocal connections relieve their poverty, not social trust. Social protection
can weaken the relationship between social capital and poverty alleviation at the aggregate and dimensional levels, reducing the
“elite capture phenomenon” of unequal social capital and bringing about equal poverty alleviation.
Our study contributes to social capital and poverty alleviation literature in three ways. First, we focus on Chinese domestic
migrant workers (hereinafter referred to as migrant workers) who were severely affected by the COVID-19 pandemic and were
neglected. Previous studies have documented that social capital can alleviate rural residents' poverty (Kamarni et al., 2019; Hong
and Tisdell, 2017; Yunus et al., 2020), but there is little knowledge concerning migrant workers’ social capital and poverty state.
Secondly, previous studies show social capital can relieve monetary poverty (Heizmann and Böhnke, 2016; Lukasiewicz et al.,
2019), but few studies investigate how social capital inuences poverty at the multidimensional level (hereinafter referred to as
the poverty level). Thirdly, social protection is essential in relieving poverty (Chan and Wong, 2020). Limited studies have explored
how social capital adjusts the relationship between social capital and poverty, especially for migrant workers who are constrained
in accessing social protection. We examine the mediation effect of social protection between migrant workers’ social capital and
poverty level.
The rest of this paper is structured as follows. Section 2 presents the literature review. Data collection and methodology are
described in Section 3. The results are shown in Section 4. Section 5 concludes the paper.
2. Literature Review And Hypotheses
Social capital is intangible and reects the resources people can access to create value through interaction, cooperation and
coordination (Priyanath and Lakshika, 2020; Tahmasebi and Askaribezayeh, 2021). Previous studies have found that social
capital can alleviate poverty through various approaches, such as promoting resource sharing (Pham and Mukhopadhaya, 2021;
Weber et al., 2013), decreasing information asymmetry (Hong and Tisdell, 2017), and lowering transaction costs (Harrison et al.,
2019; Priyanath and Lakshika, 2020). Few studies concern migrant workers’ social capital and poverty state.
Compared with the rural residents living in villages with natural social capital originating from their relatives and living
community, migrant workers’ social capital is more complex. On the one hand, social capital is created by investing in the
relationships between different economic actors (Akçomak and Müller-Zick, 2018). Migrant workers have to leave their bonding
social capital in their home town physically, migrate to cities and recreate a more heterogeneous social network with people who
come from different backgrounds, such as natives, neighbours, colleagues and individuals or groups from outside the community
(Harrison et al., 2019; Zhang et al., 2017). Although the extended social capital is crucial to migrant workers’ living situation and
economic development in their home region (Mendola, 2012), it is costly and time-consuming for migrant workers to expand
social capital (Islam and Alam, 2018; Mazelis, 2015). In addition, as newcomers to cities, migrant workers’ capital decits could
trap them in poverty (Morales, 2016). On the other hand, social exclusion and social stigma from the labour market limit migrant
workers’ opportunities to get a better life (Taylor and Foster, 2015). However, if migrant workers can build social capital in the
cities, their poverty could be signicantly improved. For example, Heizmann and Böhnke (2016) found that good connections and
social integration helps reduce immigrant income poverty in Germany. Therefore, we hypothesize the following relationship:
H1: Social capital can mitigate migrant workers’ poverty level.
Although higher-paid work opportunities in cities can help migrant workers increase their income (Garip, 2012; Mendola, 2012),
institutional exclusion can deteriorate their living conditions (Jiang et al., 2018). Public or formal social protection can empower
people to reduce vulnerability, risk, and economic shocks at the macro level (Devereux and Getu, 2016). Previous studies have

Page 4/27
found that individuals have the higher social capital to get out of poverty if they can access social protection, such as education
(or professional training) (Pillai et al., 2021; Rea et al., 2013), health and safety protection (Moyce and Schenker, 2018) and
unemployment benets (Beenstock et al., 2015; Wang, 2021). For example, Osabohien et al. (2020) found that social protection,
such as building human resources and equality of public resources, is a vital strategy to reduce poverty and inequality in Africa.
Therefore, if the local government can provide public social protection, this institutional empowerment will protect them from
marginalization (Chan and Wong, 2020). However, access to social protection could be inuenced by the level of migrant workers’
social capital. For migrant workers, a lack of political commitment by the destination government excludes them from accessing
legal institutions and social and economic benets, such as unemployment security (Morales, 2016), the urban security system,
and the urban education system because of their rural identity (Singh, 2019; Wang, 2021), causing them to spiral into poverty
(Sabates-Wheeler and Waite, 2003). In addition, social capital's “dark side” could reinforce inequalities because of existing power
asymmetry. Households with higher social capital could easily capture the benets from the public good and even gain resources
that belong to other members (Cheng et al., 2022). Therefore, we hypothesize the following relationships:
H2: Migrant workers with higher social capital have a higher social protection level.
H3: Social protection can mediate the relationship between social capital and poverty level.
Like many studies, in our study, social networks, reciprocity, and social trust are included as the dimensions of social capital
(Cheng et al., 2022; Ma et al., 2020; Priyanath and Lakshika, 2020). Social networks are the carriers of social capital (Akçomak
and Müller-Zick, 2018). Individuals with broader, more varied social networks can access more heterogenous opportunities and
resources. Social networks are formed by bonding or bridging relationships, while poor people have social networks but lack
embedded resources, and have bonding relationships but lack bridging connections (Cheng et al., 2022). However, bridging
relationships have a leverage function, facilitating poor people's upward mobility through weak ties with higher-status individuals
who can provide heterogeneous information, inuence and nancial resources (Wu et al., 2019). Migrant workers break the spatial
concentration disadvantages and expand their social networks by bridging the relationship with new neighbours, colleagues, and
their families and friends working in cities. Under this condition, social capital originating from a closed network becomes more
open (Akçomak and Müller-Zick, 2018).
The social network is the deep sediment of reciprocal exchange (Toit and Neves, 2009). One main reason is that individuals can
expand their social networks through reciprocity exchanges, which is a way to exchange resources and strengthen the mutual
trust (Islam and Alam, 2018). Reciprocity can be specic (balanced) or generalized (unbalanced), and happen in the short-term
(simultaneously) or long-term (Putnam et al., 1993). Migrant workers need to build potential benecial relationships by developing
generalized reciprocity with the people surrounding them (Rea et al., 2013). This continuing relationship of exchange is
unbalanced at any given time (Putnam et al., 1993) and is costly and time-consuming (Islam and Alam, 2018; Mazelis, 2015).
Trust comes from social participation and is the foundation of reciprocity. Trust can be divided into individual and institutional
levels (Karlina et al., 2019). This study focuses on interpersonal mutual trust. Trust is essential to promoting group decision-
making, and interpersonal exchange and reducing opportunistic behaviour (Kamarni et al., 2019; Lukasiewicz et al., 2019;
Priyanath and Lakshika, 2020). A higher trust level can improve household well-being (Ma et al., 2020). For migrant workers,
building interpersonal trust with their neighbours and colleagues in the cities can help them integrate into the community and
workplace (Rea et al., 2013).
Therefore, we hypothesize the following relationships:
H4a: Social networks can mitigate migrant workers’ poverty level.
H4b: Reciprocity level can mitigate migrant workers’ poverty level.
H4c: Social trust can mitigate migrant workers’ poverty level.
“It’s not what you know; it’s who you know” (Woolcock and Narayan, 2000, p225). This conventional aphorism sums up the
importance of social networks. Social networks are a collection of social connections and embedded resources (Lukasiewicz et

Page 5/27
al., 2019). The resources, including information, knowledge, capacity, money, social support and ow or exchange in or between
social networks, reduce the negative effect of life events (Urquhart et al., 2008). However, accessible resources depend more on
members' power position, network location, and social hierarchy (Bourdieu, 1986), while non-members could be excluded from
social networks to show loyalty to the community. To gain membership, close contact or bonding connections are critical
(Woolcock and Narayan, 2000; Yunus et al., 2020). More often, the inuential group takes advantage of its strong social network
to gain resources for group members to promote private or sectarian interests (MacGillivray, 2018). Inclusive social support from
government or non-government organizations is critical to relieve this problem. Under inclusive programmes, households isolated
from mainstream social networks also have a chance to improve their well-being (Davis and Martinez, 2014), which promotes
inclusive nance poverty alleviation, or microcredit programmes in developing countries (Han et al., 2019; Sarker and Islam,
2014).
Trust and reciprocity are rooted in bonding and bridging connections, and they are formed through interactions between
communities and formal institutions (MacGillivray, 2018). Households can gain resources through trust and reciprocal
relationships such as nancial, social, or political support (Priyanath and Lakshika, 2020). Through this process, trust and
reciprocity can promote collective actions and realize mutually benecial outcomes at a low transaction cost (Putnam et al.,
1993). However, social protection is a reciprocal process involving resource exchange, and trust plays a critical role in facilitating
the progress of resource exchange (Dankyi et al., 2017), indicating the “dark side” of reciprocity, such as clientelism, cronyism,
and nepotism (MacGillivray, 2018). People with high trust and deep reciprocity could benet from a social protection programme
to damage the poverty alleviation effect. For migrant workers, their weak social networks and low reciprocity relationships could
hinder them from enjoying those social services, and their lower social trust level would not let them strive for social protection
(Boccagni, 2011). Therefore, the accessed social protection level could be inuenced by migrant workers’ social network,
reciprocity, and social trust, which would enable them to improve poverty.
Hence, we hypothesize the following relationships:
H5a: Social protection positively mediates the relationship between social networks and poverty level.
H5b: Social protection positively mediates the relationship between reciprocity and poverty level.
H5c: Social protection positively mediates the relationship between social trust and poverty level.
3. Data And Methodology
3.1 Data collection
This study focuses on domestic migrant workers in China. The survey collected information about migrant workers’ personal and
household characteristics (e.g., age, gender, education, health condition, family size, number of earners, number of children, family
income and expenditure), their working conditions (e.g., employment status, work location, work industry, migrant work years),
social capital, social protection, and the effects of COVID-19. The survey was made available as a free online questionnaire
application (Tencent Wenjuan®) and administered online using the snowballing method to collect information about migrant
workers from June to September 2022. The current COVID-19 environment made it impossible for us to conduct face-to-face
interviews. First, we contacted the migrant workers based on the author’s social network. Next, we asked them to complete the
online questionnaire and forward it to other migrant workers around them.
A total of 1101 questionnaires were administered and returned. All questionnaires were screened by the household registration
type and migrant work years; incomplete questionnaires and questionnaires with unusual answers were excluded. Finally, 610
valid questionnaires were used in the analysis. To avoid the data collection bias from standard method variance (CMW), we allow
one IP to submit only once through a program setting in Tencent Wenjuan. In addition, for possible non-response bias, we divide
the data into two parts based on the data collecting date and use the independent samples t-test to test the difference in social
capital, poverty, income level, and expenditure level between the two data sub-sets (see Supplementary Table A1). The results
show no signicant difference between the two parts of the sample. Therefore, there is no non-response bias.

Page 6/27
3.2 Scales and measurement
The variables’ scales and measurements include the following:
Poverty level. Poverty level is the dependent variable in our study and is measured from the multidimensional perspective. Based
on the MPI Questionnaire proposed by the Oxford Poverty and Human Development Initiative (OPHI), some scholars have
developed multidimensional poverty indicators based on the living conditions in the target area (Chen et al., 2019; Pham and
Mukhopadhaya, 2021), such as life quality, economic condition, abilities and development opportunities. Considering the migrant
workers’ characteristics, we measure migrant workers’ multidimensional poverty level from three dimensions: economic status,
abilities, and social inclusion. The economic status reects migrant workers’ life quality, measured by household income and
expenditure during COVID-19. Abilities are measured by migrant workers’ education level and job stability, indicating migrant
workers’ development potential. Social inclusion suggests the social exclusion migrant workers suffer in cities and is measured
by the self-reported excluded feeling of migrant workers (see Table1). Based on these three dimensions, we use the entropy
method to weigh each dimension and calculate the value of poverty at the multidimensional level (Deutsch and Silber, 2005).

Page 7/27
Table 1
Key variables and descriptions
Variable Description Measurement
Key variable  
Poverty
level (MDP) MDP1: Economic status MDP11: Per capita of household annual income level. 5 = lower than 3,000 yuan,
4 = 3,001–10,000 yuan, 3 = 10,001–16,001 yuan, 2 = 16,000–25,000, 1 = More
than 25,000 yuan.
  MDP12: Household annual expenditure. Same scale as annual household
income.
 MDP2: Capabilities MDP21: Migrant workers’ highest education level. 5 = Middle school or lower, 4 =
High school or vocational training, 3 = College, 2 = Undergraduate, 1 =
Postgraduate or above.
  MDP22: Migrant workers’ working stability in the city. 3 = very stable, 2 = unstable,
1 = very unstable,
 MDP3: Social inclusions MDP31: I felt included and accepted in public services. A score of 1–5 refers to
strongly agree to strongly disagree.
  MDP32: I felt included and accepted in my workplace. A score of 1–5 refers to
strongly agree to strongly disagree.
  MDP33: I felt included and accepted in the neighbourhood where I stayed. A
score of 1–5 refers to strongly agree to strongly disagree.
  MDP34: I felt included and accepted in social/communal activities. A score of 1–
5 refers to strongly agree to strongly disagree.
Social
capital SC1: Social network SC11: I have many close friends in this city who can help me if needed.
  SC12: I have many close relatives in this city who can help me if needed.
 SC2: Reciprocity level SC21: I have received emotional support from family, relatives or friends when
needed.
  SC22: I have received tangible support (e.g., nancial, practical) from family,
relatives or friends when needed.
  SC23: I am (or “have been”) there to listen to others’ problems when needed.
  SC24: I have helped others with emotional or nancial support.
  SC25: If I am in trouble, people are willing to share their knowledge and
experience with me.
 SC3: Social trust SC31: I think most people would try to take advantage of me if they got a chance.
SC32: I don’t need to be too careful in dealing with people.
Social
protection The supportive policies
from the local
government of the
working city.
SS1: My kids can get an education in this city.
  SS2: I can enjoy the medical security in this city.
  SS3: I can get a loan in this city if I want to apply.
  SS4: My medical insurance can get reimbursed in this city.
Instrumental variable
Living with
families Dummy variable
representing the family
connection of the
respondent.
1 = migrant worker lives with families in the city, 0 otherwise.

Page 8/27
Variable Description Measurement
COVID-19’
impact on
social
activities
A 5-point Likert scale
representing COVID-19’s
impact on social
activities.
1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree
COVID-19’
impact on
family
conict
A 5-point Likert scale
representing COVID-19’s
impact on family
conicts.
1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree
COVID-19’s
impact on
daily life
A 5-point Likert scale
representing COVID-19’s
impact on daily life.
1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree
House type Categorical variable
representing the life
quality of the
respondent.
1 = Cheap rental house/ Public rental house, 2 = Public house provided by the
work unit, 3 = Friends' or relatives' house, 4 = Commercial house rentable in the
market, 5 = owned by themselves or family members, 6 = Other(s).
Control variable
Age Categorical variable
representing the age of
the respondent.
1 = Below 25 years old, 2 = 25–35 years old, 3 = 36–45 years old, 4 = 46–55 years
old, 5 = 56–65 years old, 6 = Above 65 years old
Gender Dummy variable
representing the gender
of the respondent.
1 = male, 2 = female.
Health Categorical variable
representing the health
condition of the
respondent.
1 = Very unhealthy, 2 = Unhealthy, 3 = Not bad, 4 = Healthy, 5 = Very healthy
Household
head Dummy variable
representing the
respondent’s role in the
household.
1 = migrant worker is the household head, 0 otherwise.
Family size Nominal variable
representing the family
size in the household.
The number of families.
Earners Nominal variable
representing the number
of earners in the
household.
The number of earners.
Dependency
ratio Nominal variable
representing the
dependency pressure in
the household.
The percentage of children in family members.
Living
facility Dummy variable
representing the life
quality of the
respondent.
1 = migrant worker has enough living facilities in the city, 0 otherwise.
Work
location Categorical variable
representing the work
location of the
respondent.
1 = north of China, 2 = coastal area of China, 3 = south of China.
Hukou
location Categorical variable
representing the hukou
location of the
respondent.
1 = north of China, 2 = coastal area of China, 3 = south of China.

Page 9/27
Social capital. This study follows previous related studies about social capital (Islam and Alam, 2018; Priyanath and Lakshika,
2020) to form the migrant worker’s social capital indicator, including social networks, reciprocity relationships, and social trust in
the cities (see Table1). Specically, we use social participation and linkages to measure migrant workers’ social networks in
cities; social help, caring, and information sharing to measure their reciprocity level; and interpersonal trust level to measure their
social trust in cities. Combining principal component factor analysis (PCA) with Cronbach’s alpha, the indicators passed the
validity and reliability tests (see Supplementary Table A2). Following Cheng et al. (2022), we use the component score and a
regression method to calculate the social capital index (SCI) (see Eq.(1)), where: is the number of retained factors; is the
variance contribution of the factor , and is the factor score of factor .
1
Social protection. Based on the main institutional constraints of Chinese domestic migrant workers in cities, we use access to
local education, medical insurance, local micro-loans, and profession train as the four indicators to measure migrant workers’
social protection level. The recipients are required to score those four items from 1–5 (from “strongly disagree” to “strongly
agree”). We use the average value of the four items to evaluate the social protection level of migrant workers obtained in the
cities.
Instrumental variables. To address the concern of endogeneity, we used instrumental variable estimation. We employ living with
families as the instrumental variable for the aggregate relationship between SCI and poverty level. Family members living in the
same community could bring about tighter social connections with their neighbours (Gertler et al., 2006), and family members,
especially wives, have more interactions with local people than migrant workers (Ha & Seong-Kyu, 2015). At the same time, living
with families does not signicantly inuence migrant workers’ poverty level. Five instrumental variables are considered for the
relationship between the dimensions of social capital and poverty level. Migrant workers’ social capital comes from their daily
interaction with their neighbours, communities, and colleagues (Chen et al., 2011). Therefore, in addition to living with families,
the impact of COVID-19 on social activities, family conict, and daily life are also considered instrumental variables. Further, the
living community is a kind of exclusive organization with its own shared beliefs, norms, and trusts (Islam and Alam, 2018; Kairiza
et al., 2021), and indicates the potential resources embedded in the social network that migrant workers can access and use
(Seferiadis et al., 2015). Therefore, the living community is associated with migrant workers’ social capital but does not connect
with their poverty level. The result reveals the instrumental variables are valid (see section 4).
Control variables. To control the relationship between social capital and the poverty level, we control for some individual
characteristics, such as age, gender, work industry, and work location of the migrant workers (Cheng et al., 2022; Islam and Alam,
2018; Pham and Mukhopadhaya, 2021).
3.3 Research method and empirical framework
To examine whether social capital can alleviate migrant workers’ poverty level, we use Eq.(2) to analyze the impact of social
capital on the poverty level.
2
The mediation effect exists when the impact of a predictor on the dependent variable is indirect through a mediator (MacKinnon
et al., 2007). We explore the role of social protection between the association of social capital and poverty. Eq.(3) is specied as:
3
n λi
i fii
SCI
=
(∑
n
i
=1
λifi
)
1
∑
n
i
=1
λi
MDP
=
α
0+
α
1
SCI
+
αiXi
+
ϵ
1
MDP
=
γ
0+
γ
1
SCI
+
γ
2
SP
+
γiXi
+
ϵ
3

Page 10/27
By comparing coecient in equations (2) and (3), we can identify the mediation role of social protection. In addition, we use
the bootstrapping approach and Sobel test to identify the indirect effect’s extent of social protection on the poverty level.
To untangle the relationship between social capital and the poverty level, we examine the relationship of each dimension of social
capital with the poverty level and the mediation effect of social protection between them based on equations (2) and (3).
Considering the three dimensions of social capital could correlate with each other (Castro and Roldán, 2013), we put those three
dimensions into the same equation:
4
On this basis, we explore whether social protection mediates each dimension of social capital with the poverty level, where:
MDP
is poverty at the multidimensional level after the COVID-19 outbreak; refers to the social capital index of migrant workers,
and , , and refer to the three dimensions of social capital: social network, reciprocity, and social trust, respectively;
is the social protection that migrant workers can access by working in cities; represents the control variables; , , , and are
coecients, and is the error term.
5
Figure 1 shows the research framework. We rst examine the integrated relationship between migrant workers’ social capital and
poverty at the aggregate level (H1, H2, and H3). Next, we split social capital into three dimensions and explored the relationships
between the social capital dimensions and poverty level (H4 and H5).
4. Results And Discussion
4.1 Statistics description
Table2 gives the descriptive statistics of the main, instrumental, and control variables. The results show that the average age of
migrant workers in our sample is 36–45 years old, and 64.8% of the migrant workers are male. Their self-reported average health
condition score is 3.85. Nearly half of the migrant workers are household heads, and over half live with their families.
Approximately 89.5% of migrant workers have adequate life facilities in cities. However, most live in a public house provided by
the work unit (36.1%) or a cheap rental house offered by the government (26.7%). The migrant workers’ dependency ratio is
28.4%, with an average of 4.42 family members per household. However, only a quarter of family members have income. Over
44% of migrant workers are working full-time. Their jobs mainly concentrate on manufacturing (19%), construction (12.6%) and
the wholesale industry (11.8%), and over half of the migrant workers were born and work on China’s coast. The average score of
social protection is 3.623, which means migrant workers have a high social protection level in cities in social security, medical
services, children's education, and loans.
β
1
MDP
=
β
0+
β
1
SN
+
β
2
SR
+
β
3
ST
+
βiXi
+
ϵ
2
SCI
SN SR ST SP
Xiα β γ θ
ϵ
MDP
=
θ
0+
θ
1
SN
+
θ
2
SR
+
θ
3
ST
+
θ
5
SP
+
βiXi
+
ϵ
4

Page 11/27
Table 2
Descriptive statistics of the study variables
Variable N Mean Std. Dev. Min Max
Main variable     
MDP 610 1.394 0.457 0.000 2.740
SCI 610 0.000 1.160 -4.884 2.215
Social network 610 0.000 1.277 -4.175 2.306
Reciprocity 610 0.000 1.684 -6.885 2.679
Social trust 610 0.000 1.217 -2.935 2.355
Mediation variable     
Social protection 610 3.623 0.872 1 5
Instrumental variable     
Living with families 610 0.579 0.494 0 1
COVID-19’s impact on social activities 610 3.820 0.971 1 5
COVID-19’s impact on family conict 610 3.064 1.235 1 5
COVID-19’s impact on daily life 610 3.756 1.011 1 5
House type     
Cheap rental houses provided by the government 610 0.267 0.443 0 1
Public house provided by the work unit 610 0.361 0.481 0 1
Friend’s or relative’s house 610 0.067 0.251 0 1
Commercial house rent in the market 610 0.162 0.369 0 1
Property owned by the family member 610 0.128 0.334 0 1
Other kinds of house 610 0.015 0.121 0 1
Control variables     
Age 610 2.308 0.842 1 6
Gender 610 0.648 0.478 0 1
Household head 610 3.848 0.928 1 5
Health condition 610 0.489 0.500 0 1
Dependency ratio 610 0.284 0.209 0 1
Earners 610 2.541 0.969 0 6
Family size 610 4.420 1.384 1 10
Living facility 610 0.895 0.307 0 1
Work industry     
Agriculture or husbandry 610 0.046 0.209 0 1
Mining 610 0.041 0.198 0 1
Manufacturing 610 0.190 0.393 0 1

Page 12/27
Variable N Mean Std. Dev. Min Max
Construction 610 0.126 0.332 0 1
Logistics 610 0.098 0.298 0 1
Wholesale and retail 610 0.118 0.323 0 1
Accommodation and catering 610 0.090 0.287 0 1
Financial services 610 0.082 0.275 0 1
Entertainment 610 0.030 0.169 0 1
Others 610 0.179 0.383 0 1
Work location     
North of China 610 0.189 0.391 0 1
Coast of China 610 0.566 0.496 0 1
South of China 610 0.246 0.431 0 1
Hukou location     
North of China 610 0.075 0.264 0 1
Coast of China 610 0.733 0.443 0 1
South of China 610 0.192 0.394 0 1
Figure 2 shows the migrant workers’ self-reported scores on the impact of COVID-19. The gure indicates COVID-19 has more
signicant impacts on migrant workers’ work, social activities, and daily life than on their family nance, children's care, and
family conict.
4.2 Aggregate results
4.2.1 The relationship between social capital and poverty
Table3 presents the impact of migrant workers’ social capital on their poverty level. Model (1) in Table3 shows migrant workers'
social capital positively inuences their poverty level. At the demographic level, migrant workers’ working industries signicantly
inuence their poverty. However, their age, gender, health condition, hukou location, work location, and whether they are household
heads do not affect their poverty level. At the household level, the household dependency ratio does not inuence migrant
workers’ poverty level, and more family members cannot help migrant workers relieve their poverty level but has the opposite
effect. Only when households of migrant workers have more income earners, their poverty level can be reduced. Although house
type does not affect migrant workers’ poverty level, if migrant workers have sucient living facilities in their home, their poverty
level is lower by 16.2%.

Page 13/27
Table 3
The effects of migrant workers’ social capital on multidimensional poverty level
(1) (2) (3) (4)
 OLS 2SRI_First stage 2SRI_Second stage 2SLS
 MDP SCI MDP MDP
Key variable    
SCI -0.148*** -0.206*** -0.206***
 (-9.520)  (-3.403) (-3.362)
Instrumental variable    
Living with families -0.029 0.510***  
 (-0.922) (5.671)  
Control variable    
Age 0.034 0.081 0.038*0.038*
 (1.528) (1.404) (1.660) (1.871)
Gender -0.015 -0.022 -0.016 -0.016
 (-0.431) (-0.222) (-0.468) (-0.459)
Household head -0.023 0.292*** -0.006 -0.006
 (-0.646) (3.183) (-0.156) (-0.150)
Health condition -0.026 0.241*** -0.012 -0.012
 (-1.533) (4.800) (-0.561) (-0.553)
Family size 0.109*** -0.075** 0.104*** 0.104***
 (7.952) (-1.994) (7.452) (7.960)
Dependency ratio -0.096 -0.329 -0.115 -0.115
 (-1.122) (-1.347) (-1.334) (-1.404)
Earners -0.066*** 0.092 -0.061*** -0.061***
 (-3.323) (1.474) (-2.983) (-3.096)
Living facility -0.162*** 0.876*** -0.111 -0.111
 (-3.236) (4.736) (-1.422) (-1.413)
Work on the Coast of China -0.024 0.274** -0.008 -0.008
 (-0.570) (2.578) (-0.186) (-0.178)
Work in the South of China 0.015 0.150 0.024 0.024
 (0.309) (1.117) (0.482) (0.470)
Hukou on the Coast of China 0.032 0.063 0.036 0.036
 (0.521) (0.397) (0.575) (0.597)
Hukou in the South of China 0.023 -0.161 0.013 0.013

Page 14/27
(1) (2) (3) (4)
 (0.327) (-0.851) (0.193) (0.190)
Mining -0.186*0.491 -0.158 -0.158
 (-1.745) (1.389) (-1.444) (-1.465)
Manufacturing -0.258*** 0.247 -0.243*** -0.243***
 (-3.109) (0.905) (-2.916) (-3.047)
Construction -0.237*** 0.011 -0.237*** -0.237***
 (-2.767) (0.038) (-2.762) (-2.877)
Logistics -0.173*-0.054 -0.176*-0.176**
 (-1.904) (-0.190) (-1.930) (-2.039)
Wholesale and retail -0.350*** 0.185 -0.339*** -0.339***
 (-3.955) (0.657) (-3.829) (-4.026)
Accommodation and catering -0.171*0.162 -0.161*-0.161*
 (-1.785) (0.563) (-1.690) (-1.830)
Financial services -0.379*** 0.242 -0.365*** -0.365***
 (-4.273) (0.795) (-4.108) (-4.040)
Entertainment -0.117 0.315 -0.099 -0.099
 (-0.967) (0.872) (-0.811) (-0.862)
Other industry -0.258*** 0.117 -0.251*** -0.251***
 (-3.075) (0.411) (-3.007) (-3.128)
Residual   0.058 
   (0.922) 
Constant 1.537*** -2.486*** 1.393*** 1.393***
 (11.176) (-5.363) (7.090) (7.095)
N
610 610 610 610
R
20.367 0.243 0.367 0.350
F-Stat
   10.67
Anderson or Kleibergen-Paap statistic
   33.65
Robust standard errors in the Model (1)-(3) in parentheses. Standard errors in parentheses in Model (4) in parentheses. *** p <
0.01, ** p < 0.05, * p < 0.1.
Since social capital and poverty are jointly determined (Grootaert et al., 2004; Yusuf, 2008), the relationship between social capital
and poverty level could have an endogeneity problem. We use the 2SRI approach and 2SLS to address endogeneity issues.
Table3 shows that living with families is a valid instrumental variable (see Models (1) to (3)), and the 2SLS (Model (4)) gives
similar results. In addition, the F value of the Anderson or Kleibergen-Paap statistic is 33.65, showing no under-identication
problem in this model. The results indicate a robust relationship between social capital and poverty level. Therefore, hypothesis

Page 15/27
(1) is supported. This result identies that social capital can mitigate migrant workers’ poverty at the multidimensional level,
aligning with the previous studies that social capital can mitigate rural residents’ poverty level (Kamarni et al., 2019; Pillai et al.,
2021; Priyanath and Lakshika, 2020).
To further test the robustness of the relationship between migrant workers’ social capital and their poverty level, we use the
entropy weighting method to calculate the value of SCI and use PCA to predict the value of the poverty level. Table4 shows that
migrant workers’ social capital strongly inuences their poverty level regardless of the measurement for SCI or poverty level. This
result offers a strong connection between social capital and poverty level. To avoid the endogeneity problem, we use the same
instrument variable for Models (2) and (4); the results presented show the inuence of migrant workers’ social capital on their
poverty level is robust.
Table 4
Robustness test: Effects of migrant workers’ social capital on poverty level
Original
model Change the measurement of
MDP Change the measurement of social
capital
 (1) (2) (3) (4) (5)
 OLS OLS 2SLS OLS 2SLS
 MDP MDP_new MDP_new MDP MDP
Key variable     
SCI -0.148*** -0.481*** -0.617***  
 (-9.520) (-18.497) (-5.795)  
SCI_new    -0.294*** -0.348***
    (-8.171) (-3.348)
Instrumental variable     
Living with families -0.029 -0.070  -0.016 
 (-0.922) (-1.273)  (-0.500) 
Control variable Controlled
Constant 1.537*** 0.457** 0.117 2.166*** 2.215***
 (11.176) (1.999) (0.342) (14.869) (13.120)
N
610 610 610 610 610
R
20.367 0.537 0.515 0.348 0.345
F-Stat
  14.30  10.57
Anderson or Kleibergen-Paap
statistic
  33.65  54.131
Robust standard errors in parentheses in Models (1), (2), and (4). Standard errors in parentheses in Models (3) and (5). *** p <
0.01, ** p < 0.05, * p < 0.1
4.2.2 The role of social protection between social capital and poverty
Table5 presents the results of the mediation effect of social protection on the impact of migrant workers’ social capital on their
poverty level. The mediation effect is estimated by the stepwise regression method and is identied by the bootstrap and Sobel–
Goodman estimations.

Page 16/27
Table 5
Mediation effect of social protection between migrant workers’ social capital
and poverty level
(1) (2) (3) (4)
 MDP Social protection MDP MDP
Key variable    
SCI -0.151*** 0.509*** -0.092***
 (-10.061) (22.857)  (-4.370)
Medication variable    
Social protection   -0.201*** -0.116***
   (-10.009) (-4.244)
Control variables Controlled
Constant 1.521*** 3.333*** 2.305*** 1.908***
 (11.150) (14.674) (15.633) (11.960)
N
610 610 610 610
R
20.366 0.586 0.363 0.386
Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1
Model (2) in Table5 shows migrant workers’ social capital signicantly inuences the social protection they can access. With a
unit increase in migrant workers’ social capital index, the average value of social protection that migrant workers can gain
increases by 0.59. In addition, with a unit increase in the average value of social protection, migrant workers’ poverty level can be
mitigated by 20.1% (see Model (3), Table5). Comparing Model (1) with (4), under the mediation effect of social protection, a unit
increase in a migrant worker’s social capital index sees their poverty level decrease by 9.2%.
To avoid potential bias from the point estimation in the mediation effect test, we adopt the bootstrapping approach to identify the
mediation effect of social protection between migrant workers’ social capital and poverty level. It shows the indirect effect of the
mediator of social protection is signicant (see Supplementary Table B1). Further, we use the Sobel–Goodman estimation
(hereinafter referred to as the Sobel test) to do a further test. Table6 shows the mediation effect of social protection accounts for
approximately 39.2% of the total effect of social capital on the poverty level. Based on this result, hypotheses (2) and (3) are
supported.
Table 6
Sobel Tests: Mediation effect of social protection of social capital on poverty
Coef Std Err Z P>|Z|
Sobel -0.059 0.014 -4.328 0.000
Goodman-1 (Aroian) -0.059 0.014 -4.324 0.000
Goodman-2 -0.059 0.014 -4.332 0.000
Proportion of total effect that is mediated: 0.392   
Ratio of indirect to direct effect: 0.644   
Ratio of total to direct effect: 1.644   

Page 17/27
This result is consistent with our expectation that migrant workers with higher social capital have a higher level of social
protection. This nding adds more evidence to previous studies that show social capital has a “dark side” and can bring about the
“elite capture” phenomenon (Cheng et al., 2022; Panda, 2015) - people with higher social capital can crowd out other people in
public goods. In addition, the result shows that migrant workers’ access to social protection can reduce their poverty level, and
public social protection plays a greater role in mitigating migrant workers’ poverty than their social capital. Our study
demonstrates that social protection can weaken the relationship between social capital and poverty level. Consistent with the
previous studies (Beenstock et al., 2015; Chan and Wong, 2020; Wang, 2021), this result highlights the importance of public policy
in reducing inequality. For migrant workers in China, the household registration system, informal work, and prejudice from local
people limit their children’s rights in education, social security and medical services (Wang, 2021). Legal public social protection
can mitigate this situation by empowering migrant workers to enjoy social benets (Beenstock et al., 2015; Wang, 2021). These
ndings illustrate that if the government provides migrant workers with access to more social protection services, the “elite
capture” phenomenon can be relieved.
4.3 The dimensional results
4.3.1 The relationship between the dimensions of social capital and
poverty
Based on the supported relationship between social capital and poverty level, we examine the relationship between the
dimensions of social capital (social networks, reciprocity, and social trust) of migrant workers and their poverty level. Table7
shows that the inuence of migrant workers’ social capital on their poverty comes from their social network and the reciprocity
levels, not from social trust in cities (see Model (2)).

Page 18/27
Table 7
The relationships between the dimensions of social capital and poverty level
(1) (2) (3) (4) (5) (6) (7) (8) (9)
 OLS OLS 2SRI
_First
stage
2SRI_First
stage 2SRI_First
stage 2SRI_Second
stage 2SLS GMM LIML
 MDP MDP Social
network Reciprocity Social
trust MDP MDP MDP MDP
Key
variable         
SCI -0.148***        
 (-9.520)        
Social
network -0.059***    -0.114** -0.114** -0.112** -0.116*
  (-3.355)    (-2.015) (-1.979) (-2.015) (-1.949)
Reciprocity  -0.060***    -0.072** -0.072** -0.071** -0.072**
  (-5.202)    (-2.086) (-2.166) (-2.081) (-2.013)
Social trust  -0.021    -0.014 -0.014 -0.017 -0.013
  (-1.339)    (-0.285) (-0.294) (-0.341) (-0.250)
Instrumental variable
Living with
families -0.029 -0.006 0.620*** 0.434*** 0.262***    
 (-0.922) (-0.181) (5.978) (3.360) (2.590)    
COVID-19’s
impact on
social
activities
 -0.017 0.150** 0.247*** 0.126**    
  (-0.944) (2.581) (3.398) (2.355)    
COVID-19’s
impact on
family
conict
 -0.003 0.115*** -0.151*** 0.344***    
  (-0.199) (2.762) (-2.985) (7.882)    
COVID-19’s
impact on
daily life
 -0.009 0.078 0.436*** -0.039    
  (-0.503) (1.614) (6.403) (-0.758)    
Public
house
provided by
the work
unit
 0.056 -0.340*** -0.044 -0.064    
  (1.328) (-3.231) (-0.326) (-0.582)    
Robust standard errors in parentheses in Models (1)-(6) and (8)-(9). Standard errors in parentheses in Model (7). *** p < 0.01,
** p < 0.05, * p < 0.1

Page 19/27
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Commercial
house rent
in the
market
 0.051 -0.436*** -0.017 -0.188    
  (1.009) (-2.934) (-0.086) (-1.342)    
Property
owned by
the family
member
 0.070 -0.493*** -0.089 0.005    
  (1.309) (-2.871) (-0.406) (0.032)    
Control
variable Controlled  
Residual 1      0.055   
      (0.936)   
Residual 2      0.012   
      (0.341)   
Residual 3      -0.006   
      (-0.125)   
Constant 1.537*** 1.582*** -3.052*** -4.811*** -2.291*** 1.385*** 1.385*** 1.381*** 1.383***
 (11.176) (9.765) (-7.005) (-6.814) (-5.215) (8.780) (8.783) (9.043) (8.977)
N
610 610 610 610 610 610 610 610 610
R
20.367 0.374 0.319 0.303 0.256 0.373 0.351 0.353 0.351
F-Stat
      10.83  
Anderson or
Kleibergen-
Paap
      30.648  
Sargan
statistic
      1.678  
Robust standard errors in parentheses in Models (1)-(6) and (8)-(9). Standard errors in parentheses in Model (7). *** p < 0.01,
** p < 0.05, * p < 0.1
Again, 2SRI and 2SLS are used to address the endogeneity problem. The 2SRI results (Models (3) to (6)) and the 2SLS results
(Model (7)) show that ve instrumental variables are valid and there are no under- and over-identication problems because the F
values of the Anderson or Kleibergen-Paap and Sargan statistics are 30.648 and 1.678 (P (chi) = 0.7947), respectively (see Model
(6)). The Generalized Method of Moments (GMM) estimation (see Models (8)) and Limited Information Maximum Likelihood
(LIML) method (see Model (9)) are used to triangulate the validity of the instrument variables. Table7 shows that whichever
estimation approach is used, the instrumental variables are valid, and the relationship between the dimensions of social capital
(social networks and reciprocity) and poverty level is robust. Therefore, hypotheses (4a) and (4b) are supported.
We show migrant workers' social networks and the reciprocity relationship built in the cities help them mitigate their poverty level.
On the one hand, the social network in the cities could be more diverse and provide more heterogeneous and valuable information
and resource to the migrant workers (Morales, 2016). On the other hand, reciprocity is the approach to strengthening social
connections to foster social capital (Mazelis, 2015). The reciprocity they built in cities can give them direct help when needed,
including material and spiritual assistance (Bilecen and Sienkiewicz, 2015). However, we found that social trust does not

Page 20/27
inuence migrant workers' poverty level. This result is similar to Meng's (2022) ndings. The potential reason is social trust is an
inherent factor and the foundation of reciprocity and social networks (Akçomak and Müller-Zick, 2018).
4.3.2 The role of social protection between the dimensions of social
capital and poverty
In this section, we examine the mediation effect of social protection between each dimension of social capital and poverty level.
Model (4) in Table8 shows that migrant workers with diversied social networks, higher reciprocity levels or higher trust levels
have more access to social protection. Under the mediation of social support, the roles of social networks and reciprocity on
poverty alleviation are still signicant, and social trust remains insignicant to migrant workers’ poverty level (Model (5) in
Table8).
Table 8
Mediation effect of social protection on the dimensions of social capital and poverty
level
(1) (2) (3) (4) (5)
 MDP MDP MDP Social protection MDP
Key variable     
SCI -0.151*** -0.092***   
 (-10.061) (-4.370)   
Social network   -0.067*** 0.235*** -0.041**
   (-4.121) (7.989) (-2.354)
Reciprocity   -0.062*** 0.191*** -0.041***
   (-5.712) (9.826) (-3.439)
Social trust   -0.024 0.112*** -0.011
   (-1.608) (4.047) (-0.821)
Mediation variable
Social protection  -0.116***   -0.112***
  (-4.244)   (-3.923)
Control variable Controlled
Constant 1.521*** 1.908*** 1.510*** 3.360*** 1.886***
 (11.150) (11.960) (10.978) (14.983) (11.523)
N
610 610 610 610 610
R
20.366 0.386 0.369 0.599 0.387
Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1
The bootstrapping approach shows the indirect consequences of the mediator of social protection in the three dimensions of
social capital are signicant (see Supplementary Table B2). The results of the Sobel test (Table9) show that the indirect effect of
social protection between social networks and poverty accounts for 39.3%. Comparably, the mediation effect of social protection
is 34.3% of the total effect of reciprocity on the poverty level. It is noted that although social trust does not affect migrant workers’
poverty level, social protection still mediates the effect of social trust on the poverty level, up to 52.3%. Therefore, hypotheses (5a)
and (5b) are supported.

Page 21/27
Table 9
Sobel test results: Mediation effect of social protection between dimensions of
social capital and poverty level
Coef Std Err Z P>|Z|
Social network    
Sobel -0.026 0.007 -3.865 0.000
Goodman-1 (Aroian) -0.026 0.007 -3.849 0.000
Goodman-2 -0.026 0.007 -3.881 0.000
Proportion of total effect that is mediated: 0.393   
Ratio of indirect to direct effect: 0.647   
Ratio of total to direct effect: 1.647   
Reciprocity    
Sobel -0.021 0.005 -3.916 0.000
Goodman-1 (Aroian) -0.021 0.005 -3.903 0.000
Goodman-2 -0.021 0.005 -3.930 0.000
Proportion of total effect that is mediated: 0.343   
Ratio of indirect to direct effect: 0.523   
Ratio of total to direct effect: 1.523   
Social interaction    
Sobel -0.012 0.004 -3.288 0.001
Goodman-1 (Aroian) -0.012 0.004 -3.253 0.001
Goodman-2 -0.012 0.004 -3.325 0.001
Proportion of total effect that is mediated: 0.523   
Ratio of indirect to direct effect: 1.095   
Ratio of total to direct effect: 2.095   
The ndings show that people with greater social networks, higher reciprocity, and trust level can have more access to social
protection, where social networks play a signicant role in capturing social protection. This result is consistent with the previous
studies that more abundant social connections can provide more information or channels and more choices. For example, people
tied to the migrant network enables them to better manage their relocation to cities (Sabates-Wheeler and Waite, 2003) and have
better access to the job market (Doln and Genicot, 2010; Pericoli et al., 2015). This is also why people try to expand their social
networks or integrate them into more extensive and diversied social networks based on building trust and reciprocal
connections. In addition, these ndings reect that social protection is critical in mediating the relationship between the
dimensions of social capital (social network and reciprocity) and poverty level. Under the inuence of social protection, migrant
workers’ social networks and reciprocity level will have less impact on their poverty level. This nding conrms the role of public
social protection in promoting social equality (Bilecen and Sienkiewicz, 2015; Faist et al., 2015). It also contributes additional
evidence about social networks and resource capture and the resource-embedded attribute of social networks (Morales, 2016;
Tong et al., 2020).
5. Conclusion And Contributions

Page 22/27
Migrant work is crucial to improve the poverty level of rural residents in China. However, as newcomers, migrant workers usually
do not have the supportive social capital in cities, which could hinder them from accessing valuable resources to reduce their
poverty level. To investigate how social capital can help migrant workers’ living conditions during the COVID-19 environment and
how government support helps migrant workers to go through this diculty, this study examines the relationship between migrant
workers’ social capital and poverty level and the mediating role of social protection between them at aggregate and dimensional
levels from the multidimensional poverty perspective.
We found that migrant workers’ social capital in cities is critical to relieve their poverty. Though migrant workers’ social capital is
unequal, migrant workers with higher social capital can access more social protection. Access to social protection not only plays
an essential role in reducing migrant workers’ poverty level but also weakens the unequal inuence of social capital on poverty.
These results provide new evidence that social capital has both a good and dark side (Das, 2004; Pillai et al., 2021). We further
split social capital into three dimensions, social networks, reciprocity, and social trust, and explore the relationship between social
capital and poverty alleviation. Our ndings show that the effect of social capital on migrant workers’ poverty alleviation comes
from their social networks and reciprocal connections in the cities, not social trust. It is noted that social trust does not directly
affect the poverty level, but it can be mediated by social protection. These ndings provide more evidence for the social capital
and poverty alleviation literature.
Our ndings suggest social capital has both good and dark sides in alleviating poverty; social protection can also diminish the
dark side. Therefore, designing more inclusive but targeted social protection policies for migrant workers in different conditions is
critical. The government can increase the inclusiveness of social protection for migrant workers. For example, provide more
information channels about institutional protection for migrant workers and lower the cost of accessing government policies. In
this way, unequal social capital or the subsequent “elite capture” phenomenon can be weakened. In addition, it is helpful for
policymakers to assess the living conditions of migrant workers and provide more targeted and accurate policies. For example,
for migrant worker newcomers who do not have friends or relatives in the destination city, the government should provide more
timely information about labour markets, public house renting or provide some subsidy in nance or living appliances to migrant
workers.
Both social capital and poverty are multidimensional. This study split social capital into three dimensions and explored how each
dimension of social capital inuences migrant workers’ poverty level. Future research can split poverty into different dimensions
and explore the inner relationships between the dimensions of social capital and poverty and the potential mediation effect.
Declarations
Ethics approval:
All procedures performed in studies involving human participants were in accordance with the ethical standards of the
institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable
ethical standards. The study was approved by Lincoln University Human Ethics Committee.
References
1. Akçomak, I. S., & Müller-Zick, H. (2018). Trust and inventive activity in Europe: causal, spatial and nonlinear forces.
Annals of
Regional Science
,
60
(3), 529–568. https://doi.org/10.1007/s00168-015-0729-2
2. Beenstock, M., Ramos, R., & Surinach, J. (2015). Migration, human capital and social capital: Lessons for the EU
neighbouring countries.
International Journal of Manpower
,
36
(4), 434–440. https://doi.org/10.1108/IJM-01-2015-0001
3. Bilecen, B., & Sienkiewicz, J. (2015). Informal social protection networks of transnational social spaces.
Population, Space
and Place
,
21
(3), 227–243.
4. Boccagni, P. (2011). Migrants’ social protection as a transnational process: Public policies and emigrant initiative in the case
of Ecuador.
International Journal of Social Welfare
,
20
(3), 318–325. https://doi.org/10.1111/j.1468-2397.2010.00747.x

Page 23/27
5. Bourdieu, P. (1986). The forms of capital. in
The Sociology of Economic Life, Third Edition
, 78–92.
https://doi.org/10.4324/9780429494338
. Castro, I., & Roldán, J. L. (2013). A mediation model between dimensions of social capital.
International Business Review
,
22
(6), 1034–1050. https://doi.org/10.1016/j.ibusrev.2013.02.004
7. Chan, S. M., & Wong, H. (2020). Impact of income, deprivation and social exclusion on subjective poverty: A structural
equation model of multidimensional overty in Hong Kong.
Social Indicators Research
,
152
(3), 971–990.
https://doi.org/10.1007/s11205-020-02476-8
. Chen, K. M., Leu, C. H., & Wang, T. M. (2019). Measurement and determinants of multidimensional poverty: Evidence from
Taiwan.
Social Indicators Research
,
145
(2), 459–478. https://doi.org/10.1007/s11205-019-02118-8
9. Chen, X., Stanton, B., Kaljee, L., Fang, X., Xiong, Q., Lin, D., Zhang, L., & Li, X. (2011). Social stigma, social capital
reconstruction and rural migrants in urban China: A population health perspective.
NIH Public Access
,
70
(1), 22–32.
10. Cheng, X., Wang, J., & Chen, K. Z. (2022). Does villager social capital hinder poverty targeting? Evidence from poverty-stricken
county of Western China.
China Economic Review
,
71
(866), 101728. https://doi.org/10.1016/j.chieco.2021.101728
11. China labor bulletin. (2021). Migrant workers and their children.
China Labor Bulletin.
May 4th, 2021. Accessed February 2nd,
2021. https://clb.org.hk/content/migrant-workers-and-their-children
12. Dankyi, E., Mazzucato, V., & Manuh, T. (2017). Reciprocity in global social protection: providing care for migrants’ children.
Oxford Development Studies
,
45
(1), 80–95. https://doi.org/10.1080/13600818.2015.1124078
13. Das, R. J. (2004). Social capital and poverty of the wage-labour class: Problems with the social capital theory.
Transactions
of the Institute of British Geographers
,
29
(1), 27–45. https://doi.org/10.1111/j.0020-2754.2004.00112.x
14. Davis, E. P., & Martinez, M. S. (2014). A review of the economic theories of poverty.
National Institute of Economic and Social
Research
,
435
, 1–65.
15. Deutsch, J., & Silber, J. (2005). Measuring multidimensional poverty: An empirical comparison of various approaches.
Review
of Income and Wealth
,
51
(1), 145–174. https://doi.org/10.1111/j.1475-4991.2005.00148.x
1. Devereux, S., & Getu, M. (Eds.). (2016).
Informal and Formal Social Protection Systems in Sub-Saharan Africa
. Fountain
Publishers.
17. Doln, S., & Genicot, G. (2010). What do networks do? The role of networks on migration and “coyote” use.
Review of
Development Economics
,
14
(2), 343–359. https://doi.org/10.1111/j.1467-9361.2010.00557.x
1. Faist, T., Bilecen, B., Barglowski, K., & Sienkiewicz, J. J. (2015). Transnational social protection: Migrants’ strategies and
patterns of inequalities.
Population, Space and Place
,
21
(3), 193–202. https://doi.org/10.1002/psp.1903
19. Garip, F. (2012). Discovering diverse mechanisms of migration: The Mexico-US stream 1970-2000.
Population and
Development Review
,
38
(3), 393–433. https://doi.org/10.1111/j.1728-4457.2012.00510.x
20. Gertler, P., Levine, D. I., & Moretti, E. (2006). Is social capital the capital of the poor? The role of family and community in
helping insure living standards against health shocks.
CESifo Economic Studies
,
52
(3), 455–499.
https://doi.org/10.1093/cesifo/i012
21. Grootaert, C., Narayan, D., Jones, V. N., & Woolcock, M. (2004). Measuring social capital: An integrated questionnaire. World
Bank Working Paper (Issue 18).
22. Ha, H.-J. J., & Seong-Kyu. (2015). Social capital and assimilation of migrant workers and foreign wives in South Korea: The
case of Wongok community.
Habitat International
,
47
, 126–135. https://doi.org/10.1016/j.habitatint.2015.01.013
23. Han, J., Wang, J., & Ma, X. (2019). Effects of farmers’ participation in inclusive nance on their vulnerability to poverty:
Evidence from Qinba poverty-stricken area in China.
Emerging Markets Finance and Trade
,
55
(5), 998–1013.
https://doi.org/10.1080/1540496X.2018.1523789
24. Harrison, J. L., Montgomery, C. A., & Jeanty, P. W. (2019). A spatial, simultaneous model of social capital and poverty.
Journal
of Behavioral and Experimental Economics
,
78
(September 2018), 183–192. https://doi.org/10.1016/j.socec.2018.09.001
25. Heizmann, B., & Böhnke, P. (2016). Migrant poverty and social capital: The impact of intra- and interethnic contacts.
Research
in Social Stratication and Mobility
,
46
, 73–85. https://doi.org/10.1016/j.rssm.2016.08.006

Page 24/27
2. Islam, M. S., & Alam, K. (2018). Does social capital reduce poverty? A cross-sectional study of rural household in Bangladesh.
International Journal of Social Economics
,
45
(11), 1515–1532. https://doi.org/10.1108/IJSE-07-2017-0295
27. Jiang, J., Qian, J., & Wen, Z. (2018). Social protection for the informal sector in urban China: Institutional constraints and self-
selection behaviour.
Journal of Social Policy
,
47
(2), 335–357. https://doi.org/10.1017/S0047279417000563
2. Kairiza, T., Kembo, G., Magadzire, V., & Chigusiwa, L. (2021). Gender gap in the impact of social capital on household food
security in Zimbabwe: Does spatial proximity matter?
Review of Economics of the Household
,
20
.
https://doi.org/10.1007/s11150-021-09592-5
29. Kamarni, N., Anshori, M., & Sukmana, R. (2019). Poverty alleviation through social capital in coastal areas: Pariaman coastal
case.
Journal of Innovation in Business and Economics
,
3
(01), 1. https://doi.org/10.22219/jibe.v3i01.7561
30. Karlina, N., Rusli, B., Suryanto, & Candradewini. (2019). An analysis of social capital in empowerment community at
uninhabitable house’s renovation fund onwest bandung regency.
Humanities and Social Sciences Reviews
,
7
(3), 74–79.
https://doi.org/10.18510/hssr.2019.7311
31. Hong, L., Tisdell, C., Wang, F. (2017).
Social Capital, Poverty and its Alleviation in a Chinese Border Region: A Case Study in
the Kirghiz Prefecture, Xinjiang
(No. 63; Working Paper on Social Economics, Policy and Development., Issue 63).
32. Li, D., Zhang, J., Yang, J., Xu, Y., Lyu, R., Zhong, L., & Wang, X. (2022). Socio-economic inequalities in health service utilization
among Chinese rural migrant workers with New Cooperative Medical Scheme: a multilevel regression approach.
BMC Public
Health
,
22
(1), 1–14. https://doi.org/10.1186/s12889-022-13486-1
33. Lukasiewicz, K., Bahar, O. S., Ali, S., Gopalan, P., Parker, G., Hawkins, R., McKay, M., & Walker, R. (2019). Getting by in New York
City: Bonding, bridging and linking capital in poverty-impacted neighborhoods.
City and Community
,
18
(1), 280–301.
https://doi.org/10.1111/cico.12373
34. Ma, X., Wang, J., Zhao, L., & Han, J. (2020). The effects of social capital on farmers’ well-being in China’s undeveloped
poverty-stricken areas.
China Agricultural Economic Review
,
12
(1), 108–121. https://doi.org/10.1108/CAER-06-2016-0087
35. MacGillivray, B. H. (2018). Beyond social capital: The norms, belief systems, and agency embedded in social networks shape
resilience to climatic and geophysical hazards.
Environmental Science and Policy
,
89
(July), 116–125.
https://doi.org/10.1016/j.envsci.2018.07.014
3. MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation Analysis David.
Annual Review of Psychology
,
35
(593).
https://doi.org/10.2307/3412968
37. Mahmud, M., & Riley, E. (2022). Adapting to an aggregate shock: The impact of the Covid-19 crisis on rural households.
Review of Economics of the Household
. https://doi.org/10.1007/s11150-022-09625-7
3. Mazelis, J. M. (2015). “I Got to Try to Give Back”: How reciprocity norms in a poor people’s organization inuence members’
social capital.
Journal of Poverty
,
19
(1), 109–131. https://doi.org/10.1080/10875549.2014.979458
39. McGrath, B. (2010). Social capital in community, family, and work lives of Brazilian migrant parents in Ireland.
Community,
Work and Family
,
13
(2), 147–165. https://doi.org/10.1080/13668800903192168
40. Mendola, M. (2012). Rural out‐migration and economic development at origin: A review of the evidence.
Journal of
International Development
,
24
, 102–122. https://doi.org/10.1002/jid
41. Meng, Z. (2022). The impact of social capital on multidimensional poverty ‐‐ An empirical research based on CGSS2015
data.
World Scientic Research Journal
,
8
(7), 856–865. https://doi.org/10.6911/WSRJ.202207
42. Morales, M. C. (2016). From social capital to inequality: Migrant networks in different stages of labor incorporation.
Sociological Forum
,
31
(3), 509–530. https://doi.org/10.1111/socf.12261
43. Moyce, S. C., & Schenker, M. (2018). Migrant workers and their occupational health and safety.
Annual Review of Public
Health
,
39
, 351–365. https://doi.org/10.1146/annurev-publhealth-040617-013714
44. Osabohien, R., Matthew, O., Ohalete, P., & Osabuohien, E. (2020). Population–poverty–inequality nexus and social protection
in Africa.
Social Indicators Research
,
151
(2), 575–598. https://doi.org/10.1007/s11205-020-02381-0
45. Osei, C. D., & Zhuang, J. (2020). Rural poverty alleviation strategies and social capital link: The mediation role of women
entrepreneurship and social innovation.
SAGE Open
,
10
(2). https://doi.org/10.1177/2158244020925504

Page 25/27
4. Panda, S. (2015). Political connections and elite capture in a poverty alleviation programme in India.
Journal of Development
Studies
,
51
(1), 50–65. https://doi.org/10.1080/00220388.2014.947281
47. Pericoli, F. M., Pierucci, E., & Ventura, L. (2015). The impact of social capital on consumption insurance and income volatility
in the UK: evidence from the British Household Panel Survey.
Review of Economics of the Household
,
13
(2), 269–295.
https://doi.org/10.1007/s11150-013-9185-x
4. Pham, A. T. Q., & Mukhopadhaya, P. (2021). Multidimensionl poverty and the role of social capital in poverty alleviation
among ethnic groups in rural Vietnam: A multilevel analysis.
Social Indicators Research
,
0123456789
.
https://doi.org/10.1007/s11205-021-02747-y
49. Pillai, V., Pandey, M., & Bhatt, B. (2021). Social sustainability at the BOP through building inclusive social capital: A case
study of Drishtee. In
Sharing Economy at the Base of the Pyramid
(pp. 301–318). https://doi.org/10.1007/978-981-16-2414-
8_13
50. Portes, A. (2009). Social capital: Its origins and applications in modern sociology.
Knowledge and Social Capital
, 43–68.
https://doi.org/10.1016/b978-0-7506-7222-1.50006-4
51. Priyanath, H. M. S., & Lakshika, L. P. C. (2020). Social capital, transaction cost and livelihood success: A case of Samurdhi
community based organization in Sri Lanka.
International Journal of Management Studies and Social Science Research
,
2
(2),
69–83.
52. Putnam, R. D. (1995). Bowling Alone: The collapse and revival of American community.
Journal of Democracy
,
6
, 65–78.
https://doi.org/10.4324/9781912282319
53. Putnam, R. D., Leonardi, R., & Nonetti, R. Y. (1993).
Making democracy work: Civic traditions in modern Italy.
Princeton
University Press. https://doi.org/10.2307/2606285
54. Rea, A., Bribosia, E., Rorive, I., & Sredanovic, D. (2013). Governing diversity: Migrant integration and multiculturalism in North
America and Europe. In
Gender Diversity, Recognition and Citizenship
. https://doi.org/10.1057/9781137318879.0009
55. Sabates-Wheeler, R., & Waite, M. (2003). Migration and Social Protection: A concept paper (No. T2; Development Research
Centre on Migration, Globalisation and Poverty, Issue December). www.migrationdrc.org/publications/working_papers/WP-
T2.pdf
5. Sarker, M., & Islam, S. (2014). Social capital and access to microcredit : Evidence from rural words missing?.
Developing
Country Studies
,
4
(7), 55–62.
57. Seferiadis, A. A., Cummings, S., Zweekhorst, M. B. M., & Bunders, J. F. G. (2015). Producing social capital as a development
strategy: Implications at the micro-level.
Progress in Development Studies
,
15
(2), 170–185.
https://doi.org/10.1177/1464993414565530
5. Singh, C. (2019). Migration as a driver of changing household structures: implications for local livelihoods and adaptation.
Migration and Development
,
8
(3), 301–319. https://doi.org/10.1080/21632324.2019.1589073
59. Tahmasebi, A., & Askaribezayeh, F. (2021). Micronance and social capital formation- a social network analysis approach.
Socio-Economic Planning Sciences
,
76
(June 2019), 100978. https://doi.org/10.1016/j.seps.2020.100978
0. Taylor, A., & Foster, J. (2015). Migrant workers and the problem of social cohesion in Canada.
Journal of International
Migration and Integration
,
16
(1), 153–172. https://doi.org/10.1007/s12134-014-0323-y
1. Toit, A. du, & Neves, D. (2009). Trading on a grant: Integrating formal and informal social protection in post-apartheid migrant
networks. No. 3; BWIP Working Paper 75.
2. Tong, D., Zhang, Y., MacLachlan, I., & Li, G. (2020). Migrant housing choices from a social capital perspective: The case of
Shenzhen, China.
Habitat International
,
96
(October 2019), 102082. https://doi.org/10.1016/j.habitatint.2019.102082
3. Urquhart, C., Liyanage, S., & Kah, M. M. O. (2008). ICTs and poverty reduction: A social capital and knowledge perspective.
Journal of Information Technology
,
23
(3), 203–213. https://doi.org/10.1057/palgrave.jit.2000121
4. Wang, F. (2021). Social capital or non-human sources? A cross-context study on information source selection of migrant
farmer workers.
Journal of Information Science
, volume?, 1–17. https://doi.org/10.1177/01655515211055129
5. Wang, J. J. (2022). The labour surplus and COVID-19: the outlook for Chinese migrant low-skilled workers.
Accounting and
Finance
,
62
(1), 577–596. https://doi.org/10.1111/ac.12800

Page 26/27
. Weber, C., Wallace, J., & Tuschke, A. (2013). Social capital, social innovation and social impact.
Frontiers of Entrepreneurship
Research
,
33
(18), Vol 33 (18).
7. Woolcock, M., & Narayan, D. (2000). Social capital: Implications for development theory, research, and policy.
World Bank
Research Observer
,
15
(2), 225–249. https://doi.org/10.1093/wbro/15.2.225
. Wu, B., Liu, L., & Carter, C. J. (2019). Bridging social capital as a resource for rural revitalization in China? A survey of
community connection of university students with home villages.
Journal of Rural Studies
,
May
.
https://doi.org/10.1016/j.jrurstud.2019.05.008
9. Yunus, S., Zainal, S., Jalil, F., & Maya Aprita Sari, C. (2020). Correlation of social capital and poverty: Farmers in Aceh.
Humanities & Social Sciences Reviews
,
8
(1), 20–26. https://doi.org/10.18510/hssr.2020.813
70. Yusuf, S. A. (2008). Social capital and household welfare in Kwara State, Nigeria.
Journal of Human Ecology
,
23
(3), 219–
229. https://doi.org/10.1080/09709274.2008.11906074
71. Zhang, Y., Zhou, X., & Lei, W. (2017). Social capital and its contingent value in poverty reduction: Evidence from Western
China.
World Development
,
93
, 350–361. https://doi.org/10.1016/j.worlddev.2016.12.034
Figures
Figure 1

Page 27/27
The research framework
Figure 2
COVID-19’s impacts on migrant workers (score 1-5 refers to “strongly disagree” to “strongly agree”)
Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
Supplementaryinformation.docx